@mc-and-his-agents/loom-installer 0.1.1

package/payload/skills/loom-handoff/.loom-runtime/shared/scripts/loom_check.py

@@ -0,0 +1,4683 @@
+#!/usr/bin/env python3
+"""Minimal Loom repository mechanical self-check."""
+
+from __future__ import annotations
+
+import re
+import shutil
+import subprocess
+import sys
+import tempfile
+import unicodedata
+import json
+import os
+from collections import Counter, defaultdict
+from dataclasses import dataclass
+from pathlib import Path
+
+from fact_chain_support import inspect_fact_chain
+from governance_surface import build_governance_surface
+from loom_flow import repo_specific_requirements_payload
+from runtime_paths import repo_local_root
+
+TOP_LEVEL_DIRS = (
+    "adoption",
+    "governance",
+    "harness",
+    "skills",
+    "templates",
+)
+
+TOP_LEVEL_FILES = (
+    "AGENTS.md",
+    "LICENSE",
+    "Makefile",
+    "README.md",
+    "VISION.md",
+    "governance-design.md",
+    "harness-design.md",
+    "system-design.md",
+)
+
+AREA_READMES = (
+    "adoption/README.md",
+    "governance/README.md",
+    "harness/README.md",
+    "skills/README.md",
+    "templates/README.md",
+)
+
+CORE_DOCS = (
+    ".github/PULL_REQUEST_TEMPLATE.md",
+    ".github/workflows/loom-check.yml",
+    "governance/principles.md",
+    "governance/review-model.md",
+    "governance/maturity-and-closing.md",
+    "governance/state-machine.md",
+    "governance/truth-and-sync-boundary.md",
+    "governance/host-object-taxonomy.md",
+    "harness/work-item-contract.md",
+    "harness/fact-chain-contract.md",
+    "harness/execution-context.md",
+    "harness/execution-chain.md",
+    "harness/checkpoint-model.md",
+    "harness/workspace-model.md",
+    "harness/workspace-lifecycle.md",
+    "harness/host-action-contract.md",
+    "harness/host-lifecycle-boundary.md",
+    "harness/reconciliation-audit.md",
+    "harness/recovery-model.md",
+    "harness/review-execution.md",
+    "harness/status-surface.md",
+    "harness/automation-frontload.md",
+    "harness/merge-checkpoint.md",
+    "harness/closeout-gate.md",
+    "harness/workspace-and-purity.md",
+    "templates/spec-suite.md",
+    "templates/pull-request.md",
+    "adoption/extraction-ledger.md",
+    "adoption/landing-map.md",
+    "adoption/rationale.md",
+    "adoption/routing-and-checkpoints.md",
+    "adoption/lightweight-retrofit-default.md",
+    "adoption/repo-companion-contract.md",
+    "adoption/repo-interop-contract.md",
+    "skills/distribution-and-adapter-contract.md",
+    "skills/registry.json",
+    "skills/install-layout.json",
+    "skills/upgrade-contract.json",
+    "skills/route-matrix.md",
+    "skills/loom-init/SKILL.md",
+    "skills/loom-init/contract.json",
+    "skills/loom-init/references/input-signals.md",
+    "skills/loom-init/references/intake-signals.md",
+    "skills/loom-init/references/output-contract.md",
+    "skills/loom-review/SKILL.md",
+    "skills/loom-review/contract.json",
+    "skills/loom-review/references/input-signals.md",
+    "skills/loom-review/references/output-contract.md",
+    "templates/review-record.md",
+    "templates/scaffold/spec.md",
+    "templates/scaffold/plan.md",
+    "tools/loom_init.py",
+    "tools/loom_flow.py",
+)
+
+AUTOMATION_FRONTLOAD_TEMPLATES = (
+    "templates/spec-suite.md",
+    "templates/pull-request.md",
+)
+
+AUTOMATION_FRONTLOAD_SKILLS = (
+    "skills/README.md",
+    "skills/distribution-and-adapter-contract.md",
+    "skills/install-layout.json",
+    "skills/route-matrix.md",
+    "skills/loom-init/SKILL.md",
+    "skills/loom-init/references/input-signals.md",
+    "skills/loom-init/references/intake-signals.md",
+    "skills/loom-init/references/output-contract.md",
+)
+
+AUTOMATION_FRONTLOAD_EXECUTION_SUPPORT = (
+    "harness/work-item-contract.md",
+    "harness/execution-context.md",
+    "harness/execution-chain.md",
+    "harness/checkpoint-model.md",
+    "harness/workspace-model.md",
+    "harness/workspace-lifecycle.md",
+    "harness/recovery-model.md",
+    "harness/status-surface.md",
+    "harness/automation-frontload.md",
+    "harness/merge-checkpoint.md",
+    "harness/workspace-and-purity.md",
+)
+
+DEMO_ASSETS = (
+    "examples/new-project/.gitkeep",
+    "examples/new-project/AGENTS.md",
+    "examples/new-project/.github/PULL_REQUEST_TEMPLATE.md",
+    "examples/new-project/.loom/bootstrap/init-result.json",
+    "examples/new-project/.loom/bootstrap/manifest.json",
+    "examples/new-project/.loom/work-items/INIT-0001.md",
+    "examples/new-project/.loom/progress/INIT-0001.md",
+    "examples/new-project/.loom/reviews/INIT-0001.json",
+    "examples/new-project/.loom/status/current.md",
+    "examples/new-project/.loom/bin/loom_init.py",
+    "examples/new-project/.loom/bin/fact_chain_support.py",
+    "examples/new-project/.loom/bin/runtime_paths.py",
+    "examples/new-project/.loom/bin/runtime_state.py",
+    "examples/new-project/.loom/bin/loom_flow.py",
+    "examples/new-project/.loom/bin/loom_check.py",
+    "examples/new-project/.loom/specs/INIT-0001/spec.md",
+    "examples/new-project/.loom/specs/INIT-0001/plan.md",
+)
+
+LINK_RE = re.compile(r"!?\[[^\]]*\]\(([^)]+)\)")
+HEADING_RE = re.compile(r"^(#{1,6})\s+(.*?)(?:\s+#+\s*)?$")
+CODE_FENCE_RE = re.compile(r"^(```|~~~)")
+EXTERNAL_SCHEME_RE = re.compile(r"^[a-zA-Z][a-zA-Z0-9+.-]*:")
+
+
+@dataclass(frozen=True)
+class Failure:
+    category: str
+    detail: str
+
+
+GOVERNANCE_SURFACE_ROUTE_SKILLS = {
+    "loom-adopt",
+    "loom-resume",
+}
+
+GOVERNANCE_SURFACE_CONTRACT_SKILLS = {
+    "loom-adopt",
+    "loom-resume",
+}
+
+REVIEW_FINDING_SEVERITIES = {"warn", "block"}
+REVIEW_FINDING_DISPOSITION_STATUSES = {"accepted", "rejected", "deferred"}
+REPO_INTERFACE_AVAILABILITY = {"absent", "companion_docs_only", "incomplete", "present"}
+REPO_INTERFACE_ENFORCEMENT = {"blocking", "advisory"}
+REPO_INTEROP_AVAILABILITY = {"absent", "incomplete", "present"}
+
+
+def repo_root_from_argv(argv: list[str]) -> Path:
+    if len(argv) > 2:
+        raise SystemExit("usage: loom_check.py [repo-root]")
+    if len(argv) == 2:
+        return Path(argv[1]).expanduser().resolve()
+    hinted_root = repo_local_root(__file__)
+    if hinted_root is not None:
+        return hinted_root
+    current = Path.cwd().resolve()
+    if (current / "skills").exists() and (current / "README.md").exists():
+        return current
+    return Path(__file__).resolve().parent.parent
+
+
+def check_required_paths(root: Path, category: str, paths: tuple[str, ...]) -> list[Failure]:
+    failures: list[Failure] = []
+    for relative_path in paths:
+        if not (root / relative_path).exists():
+            failures.append(Failure(category, f"missing `{relative_path}`"))
+    return failures
+
+
+def iter_markdown_files(root: Path) -> list[Path]:
+    return sorted(path for path in root.rglob("*.md") if path.is_file())
+
+
+def split_link_target(raw_target: str) -> tuple[str, str]:
+    target = raw_target.strip()
+    if not target:
+        return "", ""
+    if target.startswith("<") and target.endswith(">"):
+        target = target[1:-1].strip()
+    if " " in target:
+        target = target.split(" ", 1)[0]
+    if "#" in target:
+        path_part, fragment = target.split("#", 1)
+        return path_part, fragment
+    return target, ""
+
+
+def markdown_links(path: Path) -> list[tuple[int, str]]:
+    results: list[tuple[int, str]] = []
+    in_code_fence = False
+    for line_no, line in enumerate(path.read_text(encoding="utf-8").splitlines(), start=1):
+        if CODE_FENCE_RE.match(line.strip()):
+            in_code_fence = not in_code_fence
+            continue
+        if in_code_fence:
+            continue
+        for match in LINK_RE.finditer(line):
+            results.append((line_no, match.group(1)))
+    return results
+
+
+def strip_inline_markdown(text: str) -> str:
+    text = re.sub(r"`([^`]*)`", r"\1", text)
+    text = re.sub(r"!\[([^\]]*)\]\([^)]+\)", r"\1", text)
+    text = re.sub(r"\[([^\]]+)\]\([^)]+\)", r"\1", text)
+    text = re.sub(r"[*_~]", "", text)
+    return text
+
+
+def github_anchor_map(path: Path, cache: dict[Path, set[str]]) -> set[str]:
+    cached = cache.get(path)
+    if cached is not None:
+        return cached
+
+    anchors: set[str] = set()
+    duplicates: Counter[str] = Counter()
+    in_code_fence = False
+    for line in path.read_text(encoding="utf-8").splitlines():
+        stripped = line.strip()
+        if CODE_FENCE_RE.match(stripped):
+            in_code_fence = not in_code_fence
+            continue
+        if in_code_fence:
+            continue
+        match = HEADING_RE.match(line)
+        if not match:
+            continue
+        base = github_slug(strip_inline_markdown(match.group(2)))
+        if not base:
+            continue
+        duplicates[base] += 1
+        anchor = base if duplicates[base] == 1 else f"{base}-{duplicates[base] - 1}"
+        anchors.add(anchor)
+
+    cache[path] = anchors
+    return anchors
+
+
+def github_slug(text: str) -> str:
+    text = unicodedata.normalize("NFKD", text).lower().strip()
+    slug_chars: list[str] = []
+    last_was_dash = False
+    for char in text:
+        if char.isspace() or char == "-":
+            if slug_chars and not last_was_dash:
+                slug_chars.append("-")
+                last_was_dash = True
+            continue
+
+        category = unicodedata.category(char)
+        if category[0] in {"L", "N"} or category == "Mn":
+            slug_chars.append(char)
+            last_was_dash = False
+
+    return "".join(slug_chars).strip("-")
+
+
+def resolve_link_target(root: Path, source_path: Path, raw_target: str) -> tuple[Path | None, str]:
+    target, fragment = split_link_target(raw_target)
+    if not target:
+        return source_path, fragment
+    if EXTERNAL_SCHEME_RE.match(target) or target.startswith("//"):
+        return None, ""
+    if target.startswith("/"):
+        return None, ""
+
+    resolved = (source_path.parent / target).resolve()
+    if resolved.exists():
+        return resolved, fragment
+    if resolved.is_dir():
+        readme = resolved / "README.md"
+        if readme.exists():
+            return readme, fragment
+    try:
+        resolved.relative_to(root)
+    except ValueError:
+        return resolved, fragment
+    return resolved, fragment
+
+
+def check_markdown_links(root: Path) -> list[Failure]:
+    failures: list[Failure] = []
+    anchor_cache: dict[Path, set[str]] = {}
+    for markdown_path in iter_markdown_files(root):
+        for line_no, raw_target in markdown_links(markdown_path):
+            resolved, fragment = resolve_link_target(root, markdown_path, raw_target)
+            if resolved is None:
+                continue
+            if not resolved.exists():
+                detail = (
+                    f"`{markdown_path.relative_to(root)}:{line_no}` -> `{raw_target}` "
+                    f"(missing `{resolved.relative_to(root) if resolved.is_absolute() and is_within(resolved, root) else resolved}`)"
+                )
+                failures.append(Failure("markdown-links", detail))
+                continue
+            if fragment and resolved.suffix.lower() == ".md":
+                anchors = github_anchor_map(resolved, anchor_cache)
+                if fragment not in anchors:
+                    detail = (
+                        f"`{markdown_path.relative_to(root)}:{line_no}` -> `{raw_target}` "
+                        f"(missing anchor `#{fragment}` in `{resolved.relative_to(root)}`)"
+                    )
+                    failures.append(Failure("markdown-links", detail))
+    return failures
+
+
+def load_json_file(path: Path) -> object:
+    with path.open(encoding="utf-8") as handle:
+        return json.load(handle)
+
+
+def run_command(
+    root: Path,
+    args: list[str],
+    cwd: Path | None = None,
+    env: dict[str, str] | None = None,
+    timeout_seconds: float | None = None,
+) -> subprocess.CompletedProcess[str]:
+    command_env = os.environ.copy()
+    for key in ("LOOM_SOURCE_REPO_ROOT", "LOOM_INSTALLED_SKILLS_ROOT", "LOOM_RUNTIME_SCENE"):
+        command_env.pop(key, None)
+    if env:
+        command_env.update(env)
+    return subprocess.run(
+        args,
+        cwd=cwd or root,
+        check=False,
+        capture_output=True,
+        text=True,
+        env=command_env,
+        timeout=timeout_seconds,
+    )
+
+
+def load_command_json(
+    root: Path,
+    args: list[str],
+    *,
+    cwd: Path | None = None,
+    env: dict[str, str] | None = None,
+    timeout_seconds: float | None = None,
+) -> tuple[dict[str, object] | None, str | None]:
+    try:
+        result = run_command(root, args, cwd=cwd, env=env, timeout_seconds=timeout_seconds)
+    except subprocess.TimeoutExpired:
+        return None, f"command timed out after {int(timeout_seconds or 0)}s"
+    if not result.stdout.strip():
+        detail = "command produced no JSON output"
+        if result.stderr.strip():
+            detail += f": {result.stderr.strip()}"
+        return None, detail
+    try:
+        payload = json.loads(result.stdout)
+    except json.JSONDecodeError as exc:
+        return None, f"invalid JSON output: {exc.msg}"
+    if not isinstance(payload, dict):
+        return None, "command output must be a JSON object"
+    return payload, None
+
+
+def load_command_json_with_retry(
+    root: Path,
+    args: list[str],
+    *,
+    cwd: Path | None = None,
+    env: dict[str, str] | None = None,
+    timeout_seconds: float | None = None,
+    retries: int = 2,
+) -> tuple[dict[str, object] | None, str | None]:
+    transient_needles = (
+        "EOF",
+        "unknown owner type",
+        "command timed out",
+        "connection reset",
+        "TLS handshake timeout",
+    )
+    last_payload: dict[str, object] | None = None
+    last_error: str | None = None
+    for _ in range(retries):
+        payload, error = load_command_json(
+            root,
+            args,
+            cwd=cwd,
+            env=env,
+            timeout_seconds=timeout_seconds,
+        )
+        if error is None:
+            return payload, None
+        last_payload = payload
+        last_error = error
+        if not any(needle in error for needle in transient_needles):
+            return payload, error
+    return last_payload, last_error
+
+
+def payload_has_github_rate_limit(payload: object) -> bool:
+    if not isinstance(payload, dict):
+        return False
+    missing_inputs = payload.get("missing_inputs")
+    if not isinstance(missing_inputs, list):
+        return False
+    return any(isinstance(item, str) and "API rate limit exceeded" in item for item in missing_inputs)
+
+
+def prepend_path_env(bin_dir: Path, extra: dict[str, str] | None = None) -> dict[str, str]:
+    env = dict(extra or {})
+    current_path = os.environ.get("PATH", "")
+    env["PATH"] = str(bin_dir) if not current_path else f"{bin_dir}:{current_path}"
+    return env
+
+
+def write_fake_codex(
+    path: Path,
+    *,
+    mode: str,
+    tracked_edit_target: str | None = None,
+) -> None:
+    if mode == "success":
+        body = """#!/usr/bin/env python3
+import json
+import pathlib
+import sys
+
+args = sys.argv[1:]
+output_path = pathlib.Path(args[args.index("-o") + 1])
+payload = {
+    "decision": "allow",
+    "summary": "Default Codex reviewer found the item ready for merge checkpoint consumption.",
+    "findings": [
+        {
+            "id": "warn-1",
+            "summary": "Keep the follow-up validation note visible in the review record.",
+            "severity": "warn",
+            "rebuttal": None,
+            "disposition": {
+                "status": "accepted",
+                "summary": "The reviewer accepts the current validation coverage."
+            },
+            "details": "This finding is advisory and should not block merge-ready."
+        }
+    ]
+}
+output_path.parent.mkdir(parents=True, exist_ok=True)
+output_path.write_text(json.dumps(payload, ensure_ascii=False, indent=2) + "\\n", encoding="utf-8")
+sys.exit(0)
+"""
+    elif mode == "schema_drift":
+        body = """#!/usr/bin/env python3
+import json
+import pathlib
+import sys
+
+args = sys.argv[1:]
+output_path = pathlib.Path(args[args.index("-o") + 1])
+payload = {
+    "decision": "allow",
+    "findings": []
+}
+output_path.parent.mkdir(parents=True, exist_ok=True)
+output_path.write_text(json.dumps(payload, ensure_ascii=False, indent=2) + "\\n", encoding="utf-8")
+sys.exit(0)
+"""
+    elif mode == "tracked_edit":
+        target = tracked_edit_target or ""
+        body = f"""#!/usr/bin/env python3
+import json
+import pathlib
+import sys
+
+args = sys.argv[1:]
+cwd = pathlib.Path(args[args.index("-C") + 1])
+output_path = pathlib.Path(args[args.index("-o") + 1])
+target = cwd / {target!r}
+target.write_text(target.read_text(encoding="utf-8") + "\\ntracked edit from fake codex\\n", encoding="utf-8")
+payload = {{
+    "decision": "block",
+    "summary": "Tracked repository content was modified during review.",
+    "findings": [
+        {{
+            "id": "block-1",
+            "summary": "Tracked repo content changed during review execution.",
+            "severity": "block",
+            "rebuttal": None,
+            "disposition": {{
+                "status": "rejected",
+                "summary": "The run must fail closed."
+            }}
+        }}
+    ]
+}}
+output_path.parent.mkdir(parents=True, exist_ok=True)
+output_path.write_text(json.dumps(payload, ensure_ascii=False, indent=2) + "\\n", encoding="utf-8")
+sys.exit(0)
+"""
+    else:
+        raise ValueError(f"unknown fake codex mode: {mode}")
+    path.write_text(body, encoding="utf-8")
+    path.chmod(0o755)
+
+
+def require_governance_surface(
+    failures: list[Failure],
+    *,
+    category: str,
+    context: str,
+    payload: dict[str, object],
+) -> None:
+    governance_surface = payload.get("governance_surface")
+    if not isinstance(governance_surface, dict):
+        failures.append(Failure(category, f"{context} must include `governance_surface` as an object"))
+        return
+
+    required_keys = (
+        "repository_mode",
+        "loom_state",
+        "carrier_summary",
+        "execution_entry",
+        "validation_entry",
+        "review_merge_surface",
+        "github_control_plane",
+        "repo_interface",
+        "repo_interop",
+        "summary",
+        "missing_inputs",
+    )
+    for key in required_keys:
+        if key not in governance_surface:
+            failures.append(Failure(category, f"{context} governance_surface must include `{key}`"))
+
+    for key in ("repository_mode", "loom_state", "execution_entry", "validation_entry", "summary"):
+        if key in governance_surface and (not isinstance(governance_surface.get(key), str) or not governance_surface.get(key)):
+            failures.append(Failure(category, f"{context} governance_surface `{key}` must be a non-empty string"))
+    if governance_surface.get("repository_mode") not in {"new", "small-existing", "complex-existing"}:
+        failures.append(Failure(category, f"{context} governance_surface `repository_mode` must stay within the stable contract"))
+    if governance_surface.get("loom_state") not in {"active", "partial", "absent"}:
+        failures.append(Failure(category, f"{context} governance_surface `loom_state` must stay within the stable contract"))
+
+    missing_inputs = governance_surface.get("missing_inputs")
+    if missing_inputs is not None and not isinstance(missing_inputs, list):
+        failures.append(Failure(category, f"{context} governance_surface `missing_inputs` must be a list"))
+
+    carrier_summary = governance_surface.get("carrier_summary")
+    if not isinstance(carrier_summary, dict):
+        failures.append(Failure(category, f"{context} governance_surface must include `carrier_summary`"))
+    else:
+        required_carriers = ("work_item", "recovery", "review", "status_surface", "spec_path", "plan_path")
+        if set(carrier_summary.keys()) != set(required_carriers):
+            failures.append(Failure(category, f"{context} governance_surface carrier keys must stay within the stable contract"))
+        for carrier in required_carriers:
+            entry = carrier_summary.get(carrier)
+            if not isinstance(entry, dict):
+                failures.append(Failure(category, f"{context} governance_surface carrier `{carrier}` must be an object"))
+                continue
+            if entry.get("status") not in {"present", "missing", "planned"}:
+                failures.append(
+                    Failure(category, f"{context} governance_surface carrier `{carrier}` status must stay within the stable contract")
+                )
+            for field in ("locator", "source"):
+                value = entry.get(field)
+                if not isinstance(value, str) or not value:
+                    failures.append(
+                        Failure(category, f"{context} governance_surface carrier `{carrier}` must include non-empty `{field}`")
+                    )
+
+    review_merge_surface = governance_surface.get("review_merge_surface")
+    if review_merge_surface is not None and not isinstance(review_merge_surface, dict):
+        failures.append(Failure(category, f"{context} governance_surface `review_merge_surface` must be an object"))
+    elif isinstance(review_merge_surface, dict):
+        for key in ("pr_template", "validation_surface", "merge_surface"):
+            value = review_merge_surface.get(key)
+            if not isinstance(value, str) or not value:
+                failures.append(Failure(category, f"{context} governance_surface `review_merge_surface.{key}` must be a non-empty string"))
+
+    github_control_plane = governance_surface.get("github_control_plane")
+    if github_control_plane is not None and not isinstance(github_control_plane, dict):
+        failures.append(Failure(category, f"{context} governance_surface `github_control_plane` must be an object"))
+    elif isinstance(github_control_plane, dict):
+        for key in ("repository", "default_branch", "branch_protection", "required_checks", "pr_reviews"):
+            if key not in github_control_plane:
+                failures.append(Failure(category, f"{context} governance_surface `github_control_plane.{key}` must exist"))
+        for key in ("repository", "default_branch"):
+            value = github_control_plane.get(key)
+            if not isinstance(value, str) or not value:
+                failures.append(Failure(category, f"{context} governance_surface `github_control_plane.{key}` must be a non-empty string"))
+        if github_control_plane.get("branch_protection") not in {"enabled", "disabled", "unknown"}:
+            failures.append(Failure(category, f"{context} governance_surface `github_control_plane.branch_protection` must stay within the stable contract"))
+        if github_control_plane.get("pr_reviews") not in {"required", "not_required", "unknown"}:
+            failures.append(Failure(category, f"{context} governance_surface `github_control_plane.pr_reviews` must stay within the stable contract"))
+        required_checks = github_control_plane.get("required_checks")
+        if not (
+            required_checks == "unknown"
+            or (isinstance(required_checks, list) and all(isinstance(item, str) and item for item in required_checks))
+        ):
+            failures.append(Failure(category, f"{context} governance_surface `github_control_plane.required_checks` must be `unknown` or a string list"))
+
+    require_repo_interface_payload(
+        failures,
+        category=category,
+        context=f"{context} governance_surface.repo_interface",
+        payload=governance_surface.get("repo_interface"),
+    )
+    require_repo_interop_payload(
+        failures,
+        category=category,
+        context=f"{context} governance_surface.repo_interop",
+        payload=governance_surface.get("repo_interop"),
+    )
+
+
+def require_locator_entry(
+    failures: list[Failure],
+    *,
+    category: str,
+    context: str,
+    payload: object,
+    allowed_statuses: set[str],
+) -> None:
+    if not isinstance(payload, dict):
+        failures.append(Failure(category, f"{context} must be an object"))
+        return
+    if payload.get("status") not in allowed_statuses:
+        failures.append(Failure(category, f"{context} status must stay within the stable contract"))
+    for field in ("locator", "source"):
+        value = payload.get(field)
+        if not isinstance(value, str) or not value:
+            failures.append(Failure(category, f"{context} must include non-empty `{field}`"))
+
+
+def require_repo_interface_payload(
+    failures: list[Failure],
+    *,
+    category: str,
+    context: str,
+    payload: object,
+) -> None:
+    if not isinstance(payload, dict):
+        failures.append(Failure(category, f"{context} must be an object"))
+        return
+    if payload.get("availability") not in REPO_INTERFACE_AVAILABILITY:
+        failures.append(Failure(category, f"{context} availability must stay within the stable contract"))
+    require_locator_entry(
+        failures,
+        category=category,
+        context=f"{context}.manifest",
+        payload=payload.get("manifest"),
+        allowed_statuses={"present", "missing"},
+    )
+    for key in ("companion_entry", "repo_specific_requirements", "specialized_gates"):
+        require_locator_entry(
+            failures,
+            category=category,
+            context=f"{context}.{key}",
+            payload=payload.get(key),
+            allowed_statuses={"present", "missing"},
+        )
+    if not isinstance(payload.get("summary"), str) or not payload.get("summary"):
+        failures.append(Failure(category, f"{context} must include non-empty `summary`"))
+    if not isinstance(payload.get("missing_inputs"), list):
+        failures.append(Failure(category, f"{context} must include `missing_inputs` as a list"))
+
+
+def require_repo_interop_payload(
+    failures: list[Failure],
+    *,
+    category: str,
+    context: str,
+    payload: object,
+) -> None:
+    if not isinstance(payload, dict):
+        failures.append(Failure(category, f"{context} must be an object"))
+        return
+    if payload.get("availability") not in REPO_INTEROP_AVAILABILITY:
+        failures.append(Failure(category, f"{context} availability must stay within the stable contract"))
+    require_locator_entry(
+        failures,
+        category=category,
+        context=f"{context}.contract",
+        payload=payload.get("contract"),
+        allowed_statuses={"present", "missing"},
+    )
+    for key in ("host_adapters", "repo_native_carriers", "shadow_surfaces"):
+        require_locator_entry(
+            failures,
+            category=category,
+            context=f"{context}.{key}",
+            payload=payload.get(key),
+            allowed_statuses={"present", "missing"},
+        )
+    if not isinstance(payload.get("summary"), str) or not payload.get("summary"):
+        failures.append(Failure(category, f"{context} must include non-empty `summary`"))
+    if not isinstance(payload.get("missing_inputs"), list):
+        failures.append(Failure(category, f"{context} must include `missing_inputs` as a list"))
+
+
+def require_repo_specific_requirements_payload(
+    failures: list[Failure],
+    *,
+    category: str,
+    context: str,
+    payload: object,
+    expected_surface: str,
+) -> None:
+    if not isinstance(payload, dict):
+        failures.append(Failure(category, f"{context} must be an object"))
+        return
+    if payload.get("surface") != expected_surface:
+        failures.append(Failure(category, f"{context} must report `surface: {expected_surface}`"))
+    if payload.get("result") not in {"pass", "block"}:
+        failures.append(Failure(category, f"{context} result must be `pass` or `block`"))
+    if not isinstance(payload.get("summary"), str) or not payload.get("summary"):
+        failures.append(Failure(category, f"{context} must include non-empty `summary`"))
+    if not isinstance(payload.get("missing_inputs"), list):
+        failures.append(Failure(category, f"{context} must include `missing_inputs`"))
+    if payload.get("fallback_to") not in {None, "build", "merge"}:
+        failures.append(Failure(category, f"{context} fallback must stay within the stable contract"))
+    for key in ("declared_requirements", "blocking_requirements", "advisory_requirements"):
+        entries = payload.get(key)
+        if not isinstance(entries, list):
+            failures.append(Failure(category, f"{context} must include `{key}` as a list"))
+            continue
+        for index, entry in enumerate(entries):
+            if not isinstance(entry, dict):
+                failures.append(Failure(category, f"{context} {key}[{index}] must be an object"))
+                continue
+            for field in ("id", "summary", "locator", "enforcement"):
+                value = entry.get(field)
+                if not isinstance(value, str) or not value:
+                    failures.append(Failure(category, f"{context} {key}[{index}] missing `{field}`"))
+            if entry.get("enforcement") not in REPO_INTERFACE_ENFORCEMENT:
+                failures.append(Failure(category, f"{context} {key}[{index}] enforcement must stay within the stable contract"))
+    declared = payload.get("declared_requirements")
+    blocking = payload.get("blocking_requirements")
+    advisory = payload.get("advisory_requirements")
+    if isinstance(declared, list) and isinstance(blocking, list) and isinstance(advisory, list):
+        if len(declared) != len(blocking) + len(advisory):
+            failures.append(Failure(category, f"{context} declared requirements must split cleanly into blocking and advisory"))
+
+
+def require_shadow_parity_payload(
+    failures: list[Failure],
+    *,
+    category: str,
+    context: str,
+    payload: object,
+    expected_reports: int,
+) -> None:
+    if not isinstance(payload, dict):
+        failures.append(Failure(category, f"{context} must be an object"))
+        return
+    if payload.get("command") != "shadow-parity":
+        failures.append(Failure(category, f"{context} must report `command: shadow-parity`"))
+    if payload.get("result") not in {"pass", "warn"}:
+        failures.append(Failure(category, f"{context} result must be `pass` or `warn`"))
+    if payload.get("fallback_to") is not None:
+        failures.append(Failure(category, f"{context} fallback_to must remain `null`"))
+    if not isinstance(payload.get("summary"), str) or not payload.get("summary"):
+        failures.append(Failure(category, f"{context} must include non-empty `summary`"))
+    if not isinstance(payload.get("missing_inputs"), list):
+        failures.append(Failure(category, f"{context} must include `missing_inputs`"))
+    require_runtime_state_payload(
+        failures,
+        category=category,
+        context=context,
+        payload=payload.get("runtime_state"),
+        expected_scene="repo-local-demo",
+        expected_carrier="repo-local-wrapper",
+        allowed_results={"pass"},
+    )
+    governance_surface = {"governance_surface": payload.get("governance_surface")}
+    if isinstance(payload.get("governance_surface"), dict):
+        require_governance_surface(
+            failures,
+            category=category,
+            context=context,
+            payload=governance_surface,
+        )
+    reports = payload.get("reports")
+    if not isinstance(reports, list):
+        failures.append(Failure(category, f"{context} must include `reports` as a list"))
+        return
+    if len(reports) != expected_reports:
+        failures.append(Failure(category, f"{context} must include {expected_reports} parity reports"))
+    for index, report in enumerate(reports):
+        if not isinstance(report, dict):
+            failures.append(Failure(category, f"{context} reports[{index}] must be an object"))
+            continue
+        if report.get("surface") not in {"admission", "review", "merge_ready", "closeout"}:
+            failures.append(Failure(category, f"{context} reports[{index}] must declare a known surface"))
+        if report.get("result") not in {"match", "mismatch", "unreadable"}:
+            failures.append(Failure(category, f"{context} reports[{index}] result must stay within the stable contract"))
+        if not isinstance(report.get("summary"), str) or not report.get("summary"):
+            failures.append(Failure(category, f"{context} reports[{index}] must include non-empty `summary`"))
+        if not isinstance(report.get("missing_inputs"), list):
+            failures.append(Failure(category, f"{context} reports[{index}] must include `missing_inputs`"))
+        for key in ("host_adapters", "repo_native_carriers"):
+            if not isinstance(report.get(key), list):
+                failures.append(Failure(category, f"{context} reports[{index}] must include `{key}` as a list"))
+        for surface_key in ("loom_surface", "repo_surface"):
+            surface_payload = report.get(surface_key)
+            if not isinstance(surface_payload, dict):
+                failures.append(Failure(category, f"{context} reports[{index}] must include `{surface_key}`"))
+                continue
+            if surface_payload.get("status") not in {"readable", "missing"}:
+                failures.append(Failure(category, f"{context} reports[{index}] `{surface_key}.status` must stay within the stable contract"))
+            locator = surface_payload.get("locator")
+            if not isinstance(locator, str) or not locator:
+                failures.append(Failure(category, f"{context} reports[{index}] `{surface_key}.locator` must be non-empty"))
+
+
+def require_host_lifecycle_payload(
+    failures: list[Failure],
+    *,
+    category: str,
+    context: str,
+    payload: dict[str, object],
+) -> None:
+    if payload.get("result") not in {"pass", "block"}:
+        failures.append(Failure(category, f"{context} must return `pass` or `block`"))
+    if payload.get("fallback_to") not in {None, "admission"}:
+        failures.append(Failure(category, f"{context} fallback must be `null` or `admission`"))
+    if not isinstance(payload.get("summary"), str) or not payload.get("summary"):
+        failures.append(Failure(category, f"{context} must include a non-empty `summary`"))
+    if not isinstance(payload.get("missing_inputs"), list):
+        failures.append(Failure(category, f"{context} must include `missing_inputs`"))
+
+    objects = payload.get("objects")
+    if not isinstance(objects, dict):
+        failures.append(Failure(category, f"{context} must include `objects`"))
+        return
+
+    workspace = objects.get("workspace")
+    branch = objects.get("branch")
+    pr = objects.get("pr")
+    worktree = objects.get("worktree")
+    for key, value in (("workspace", workspace), ("branch", branch), ("pr", pr), ("worktree", worktree)):
+        if not isinstance(value, dict):
+            failures.append(Failure(category, f"{context} must include `{key}`"))
+    if not isinstance(workspace, dict) or not isinstance(branch, dict) or not isinstance(pr, dict) or not isinstance(worktree, dict):
+        return
+
+    if workspace.get("ownership") != "loom":
+        failures.append(Failure(category, f"{context} workspace ownership must stay `loom`"))
+    for field in ("entry", "path", "lifecycle_entry"):
+        value = workspace.get(field)
+        if not isinstance(value, str) or not value:
+            failures.append(Failure(category, f"{context} workspace must include non-empty `{field}`"))
+
+    if branch.get("ownership") != "host":
+        failures.append(Failure(category, f"{context} branch ownership must stay `host`"))
+    if branch.get("purity_status") not in {"report_only", "host_managed_without_local_branch"}:
+        failures.append(Failure(category, f"{context} branch purity_status must stay within the stable contract"))
+    if not isinstance(branch.get("next_action"), str) or not branch.get("next_action"):
+        failures.append(Failure(category, f"{context} branch must include non-empty `next_action`"))
+
+    if pr.get("ownership") != "host":
+        failures.append(Failure(category, f"{context} PR ownership must stay `host`"))
+    if pr.get("purity_status") != "report_only":
+        failures.append(Failure(category, f"{context} PR purity_status must stay `report_only`"))
+    if not isinstance(pr.get("next_action"), str) or not pr.get("next_action"):
+        failures.append(Failure(category, f"{context} PR must include non-empty `next_action`"))
+
+    if worktree.get("ownership") != "host":
+        failures.append(Failure(category, f"{context} worktree ownership must stay `host`"))
+    if worktree.get("status") != "host_managed":
+        failures.append(Failure(category, f"{context} worktree status must stay `host_managed`"))
+    for field in ("cwd_within_repo", "next_action"):
+        value = worktree.get(field)
+        if not isinstance(value, str) or not value:
+            failures.append(Failure(category, f"{context} worktree must include non-empty `{field}`"))
+
+
+def require_reconciliation_payload(
+    failures: list[Failure],
+    *,
+    category: str,
+    context: str,
+    payload: object,
+) -> None:
+    if not isinstance(payload, dict):
+        failures.append(Failure(category, f"{context} must include `reconciliation` as an object"))
+        return
+    if payload.get("command") != "reconciliation":
+        failures.append(Failure(category, f"{context} must report `command: reconciliation`"))
+    if payload.get("operation") != "audit":
+        failures.append(Failure(category, f"{context} must report `operation: audit`"))
+    if payload.get("result") not in {"pass", "warn", "fix-needed", "block"}:
+        failures.append(Failure(category, f"{context} returned an unknown reconciliation result"))
+    if not isinstance(payload.get("summary"), str) or not payload.get("summary"):
+        failures.append(Failure(category, f"{context} must include a non-empty reconciliation `summary`"))
+    if not isinstance(payload.get("missing_inputs"), list):
+        failures.append(Failure(category, f"{context} must include reconciliation `missing_inputs`"))
+    if payload.get("fallback_to") not in {None, "manual-reconciliation"}:
+        failures.append(Failure(category, f"{context} reconciliation fallback must be `null` or `manual-reconciliation`"))
+    findings = payload.get("findings")
+    if not isinstance(findings, list):
+        failures.append(Failure(category, f"{context} must include reconciliation `findings` as a list"))
+        return
+    for finding in findings:
+        if not isinstance(finding, dict):
+            failures.append(Failure(category, f"{context} reconciliation findings must be JSON objects"))
+            continue
+        if finding.get("kind") not in {"absorbed_but_open", "parent_drift", "project_drift"}:
+            failures.append(Failure(category, f"{context} reconciliation finding kind must stay within the stable contract"))
+        if finding.get("severity") not in {"warn", "fix-needed", "block"}:
+            failures.append(Failure(category, f"{context} reconciliation finding severity must stay within the stable contract"))
+        if not isinstance(finding.get("subject"), str) or not finding.get("subject"):
+            failures.append(Failure(category, f"{context} reconciliation findings must include non-empty `subject`"))
+        if not isinstance(finding.get("evidence"), dict):
+            failures.append(Failure(category, f"{context} reconciliation findings must include `evidence`"))
+        if not isinstance(finding.get("recommended_action"), str) or not finding.get("recommended_action"):
+            failures.append(Failure(category, f"{context} reconciliation findings must include non-empty `recommended_action`"))
+
+
+def require_closeout_reconciliation_contract(
+    failures: list[Failure],
+    *,
+    category: str,
+    context: str,
+    payload: dict[str, object],
+) -> None:
+    reconciliation = payload.get("reconciliation")
+    if reconciliation is None:
+        return
+    require_reconciliation_payload(
+        failures,
+        category=category,
+        context=f"{context} reconciliation",
+        payload=reconciliation,
+    )
+    if not isinstance(reconciliation, dict):
+        return
+    reconciliation_result = reconciliation.get("result")
+    closeout_result = payload.get("result")
+    fallback_to = payload.get("fallback_to")
+    if reconciliation_result == "fix-needed":
+        if closeout_result != "block":
+            failures.append(Failure(category, f"{context} must block when reconciliation returns `fix-needed`"))
+        if fallback_to != "reconciliation-sync":
+            failures.append(Failure(category, f"{context} must point `fix-needed` reconciliation drift to `reconciliation-sync`"))
+    if reconciliation_result == "block":
+        if closeout_result != "block":
+            failures.append(Failure(category, f"{context} must block when reconciliation returns `block`"))
+        if fallback_to != "manual-reconciliation":
+            failures.append(Failure(category, f"{context} must point blocked reconciliation drift to `manual-reconciliation`"))
+
+
+def require_review_record_contract(
+    failures: list[Failure],
+    *,
+    category: str,
+    context: str,
+    payload: object,
+) -> None:
+    if not isinstance(payload, dict):
+        failures.append(Failure(category, f"{context} must include a review record object"))
+        return
+    findings = payload.get("findings")
+    if not isinstance(findings, list):
+        failures.append(Failure(category, f"{context} must include review `findings` as a list"))
+        return
+    for list_field in ("blocking_issues", "follow_ups"):
+        if not isinstance(payload.get(list_field), list):
+            failures.append(Failure(category, f"{context} must include review `{list_field}` as a list"))
+    consumed_inputs = payload.get("consumed_inputs")
+    if consumed_inputs is not None:
+        if not isinstance(consumed_inputs, dict):
+            failures.append(Failure(category, f"{context} review `consumed_inputs` must be an object when present"))
+        else:
+            for key in ("engine_adapter", "engine_evidence", "normalized_findings"):
+                value = consumed_inputs.get(key)
+                if value is not None and (not isinstance(value, str) or not value):
+                    failures.append(Failure(category, f"{context} review consumed input `{key}` must be null or a non-empty string"))
+    for finding in findings:
+        if not isinstance(finding, dict):
+            failures.append(Failure(category, f"{context} review findings must be JSON objects"))
+            continue
+        if not isinstance(finding.get("id"), str) or not finding.get("id"):
+            failures.append(Failure(category, f"{context} review findings must include non-empty `id`"))
+        if not isinstance(finding.get("summary"), str) or not finding.get("summary"):
+            failures.append(Failure(category, f"{context} review findings must include non-empty `summary`"))
+        if finding.get("severity") not in REVIEW_FINDING_SEVERITIES:
+            failures.append(Failure(category, f"{context} review finding severity must stay within the stable contract"))
+        rebuttal = finding.get("rebuttal")
+        if rebuttal is not None and (not isinstance(rebuttal, str) or not rebuttal):
+            failures.append(
+                Failure(category, f"{context} review finding `rebuttal` must be `null` or a non-empty string")
+            )
+        disposition = finding.get("disposition")
+        if disposition is not None:
+            if not isinstance(disposition, dict):
+                failures.append(Failure(category, f"{context} review finding disposition must be `null` or an object"))
+                continue
+            if disposition.get("status") not in REVIEW_FINDING_DISPOSITION_STATUSES:
+                failures.append(Failure(category, f"{context} review finding disposition status must stay within the stable contract"))
+            if not isinstance(disposition.get("summary"), str) or not disposition.get("summary"):
+                failures.append(Failure(category, f"{context} review finding disposition must include non-empty `summary`"))
+
+
+def require_review_run_payload(
+    failures: list[Failure],
+    *,
+    category: str,
+    context: str,
+    payload: object,
+    expected_result: set[str],
+) -> None:
+    if not isinstance(payload, dict):
+        failures.append(Failure(category, f"{context} must return a JSON object"))
+        return
+    if payload.get("command") != "review":
+        failures.append(Failure(category, f"{context} must report `command: review`"))
+    if payload.get("operation") != "run":
+        failures.append(Failure(category, f"{context} must report `operation: run`"))
+    if payload.get("result") not in expected_result:
+        failures.append(Failure(category, f"{context} returned an unexpected result"))
+    for key in ("item", "state_check", "runtime_evidence", "build_checkpoint", "review", "current_checkpoint", "engine", "manual_review"):
+        if not isinstance(payload.get(key), dict):
+            failures.append(Failure(category, f"{context} must include `{key}`"))
+    require_runtime_state_payload(
+        failures,
+        category=category,
+        context=context,
+        payload=payload.get("runtime_state"),
+        allowed_results={"pass", "block"},
+    )
+    engine = payload.get("engine")
+    if not isinstance(engine, dict):
+        return
+    if engine.get("engine") != "codex":
+        failures.append(Failure(category, f"{context} engine must stay `codex` for the default path"))
+    if engine.get("adapter") != "loom/default-codex":
+        failures.append(Failure(category, f"{context} adapter must stay `loom/default-codex`"))
+    if engine.get("result") not in {"pass", "block", "not_run"}:
+        failures.append(Failure(category, f"{context} engine result must stay within the stable contract"))
+    if engine.get("failure_reason") not in {None, "engine_unavailable", "schema_drift", "runtime_conflict", "repo_diff_detected"}:
+        failures.append(Failure(category, f"{context} engine failure reason must stay within the stable contract"))
+    evidence = engine.get("evidence")
+    if engine.get("result") == "not_run":
+        if evidence is not None:
+            failures.append(Failure(category, f"{context} engine evidence must be null when the engine is not run"))
+    else:
+        if not isinstance(evidence, dict):
+            failures.append(Failure(category, f"{context} engine must include `evidence` when it runs"))
+        else:
+            for key in ("runtime_root", "prompt", "raw_result", "normalized_findings", "metadata"):
+                value = evidence.get(key)
+                if not isinstance(value, str) or not value:
+                    failures.append(Failure(category, f"{context} engine evidence must include non-empty `{key}`"))
+    manual_review = payload.get("manual_review")
+    if isinstance(manual_review, dict):
+        if not isinstance(manual_review.get("summary"), str) or not manual_review.get("summary"):
+            failures.append(Failure(category, f"{context} manual_review must include non-empty `summary`"))
+        if not isinstance(manual_review.get("review_record_path"), str) or not manual_review.get("review_record_path"):
+            failures.append(Failure(category, f"{context} manual_review must include `review_record_path`"))
+        if manual_review.get("recommended_kind") not in {"general_review", "code_review", "spec_review"}:
+            failures.append(Failure(category, f"{context} manual_review recommended kind must stay within the stable contract"))
+        if not isinstance(manual_review.get("command"), list):
+            failures.append(Failure(category, f"{context} manual_review must include `command` as a list"))
+    review_record_input = payload.get("review_record_input")
+    if payload.get("result") == "pass":
+        if not isinstance(review_record_input, dict):
+            failures.append(Failure(category, f"{context} must include `review_record_input` when engine review passes"))
+        else:
+            for key in ("decision", "summary", "reviewer", "kind", "findings_file", "engine_adapter", "engine_evidence", "normalized_findings"):
+                value = review_record_input.get(key)
+                if not isinstance(value, str) or not value:
+                    failures.append(Failure(category, f"{context} review_record_input must include non-empty `{key}`"))
+            if review_record_input.get("decision") not in {"allow", "block", "fallback"}:
+                failures.append(Failure(category, f"{context} review_record_input decision must stay within the stable contract"))
+            if review_record_input.get("reviewer") != "loom/default-codex":
+                failures.append(Failure(category, f"{context} review_record_input reviewer must stay `loom/default-codex`"))
+
+
+def require_runtime_state_payload(
+    failures: list[Failure],
+    *,
+    category: str,
+    context: str,
+    payload: object,
+    expected_scene: str | None = None,
+    expected_carrier: str | None = None,
+    allowed_results: set[str] | None = None,
+) -> None:
+    if not isinstance(payload, dict):
+        failures.append(Failure(category, f"{context} must include `runtime_state` as an object"))
+        return
+    if payload.get("result") not in (allowed_results or {"pass", "block"}):
+        failures.append(Failure(category, f"{context} runtime_state.result must stay within the stable contract"))
+    if expected_scene is not None and payload.get("scene") != expected_scene:
+        failures.append(Failure(category, f"{context} runtime_state.scene must be `{expected_scene}`"))
+    if expected_carrier is not None and payload.get("carrier") != expected_carrier:
+        failures.append(Failure(category, f"{context} runtime_state.carrier must be `{expected_carrier}`"))
+    if payload.get("entry_family") not in {"loom-init", "loom-flow"}:
+        failures.append(Failure(category, f"{context} runtime_state.entry_family must stay within the stable contract"))
+    if not isinstance(payload.get("runtime_root"), str) or not payload.get("runtime_root"):
+        failures.append(Failure(category, f"{context} runtime_state must include non-empty `runtime_root`"))
+    checks = payload.get("checks")
+    if not isinstance(checks, dict):
+        failures.append(Failure(category, f"{context} runtime_state must include `checks`"))
+        return
+    for key in ("scene_marker", "carrier_layout", "registry_contract", "shared_runtime", "referenced_resources"):
+        check = checks.get(key)
+        if not isinstance(check, dict):
+            failures.append(Failure(category, f"{context} runtime_state must include check `{key}`"))
+            continue
+        if check.get("status") not in {"pass", "block", "not_applicable"}:
+            failures.append(Failure(category, f"{context} runtime_state check `{key}` returned an unknown status"))
+        if not isinstance(check.get("summary"), str) or not check.get("summary"):
+            failures.append(Failure(category, f"{context} runtime_state check `{key}` must include non-empty `summary`"))
+
+
+def require_route_payload(
+    failures: list[Failure],
+    *,
+    category: str,
+    context: str,
+    payload: object,
+    expected_skill: str,
+    expected_mode: str,
+    expected_runtime_scene: str | None = None,
+    expected_runtime_carrier: str | None = None,
+    allowed_results: set[str] | None = None,
+) -> None:
+    if not isinstance(payload, dict):
+        failures.append(Failure(category, f"{context} must return a JSON object"))
+        return
+    if payload.get("command") != "route":
+        failures.append(Failure(category, f"{context} must report `command: route`"))
+    if payload.get("result") not in (allowed_results or {"pass"}):
+        failures.append(Failure(category, f"{context} result must stay within the stable contract"))
+    if payload.get("selected_skill") != expected_skill:
+        failures.append(Failure(category, f"{context} must select `{expected_skill}`"))
+    if payload.get("mode") != expected_mode:
+        failures.append(Failure(category, f"{context} must report `mode: {expected_mode}`"))
+    if not isinstance(payload.get("matched_signals"), list):
+        failures.append(Failure(category, f"{context} must include `matched_signals`"))
+    if not isinstance(payload.get("missing_inputs"), list):
+        failures.append(Failure(category, f"{context} must include `missing_inputs`"))
+    if payload.get("fallback_to") not in {"loom-init", "refresh-install", "rebootstrap-runtime", "manual-runtime-reconciliation", None}:
+        failures.append(Failure(category, f"{context} fallback must stay within the stable contract"))
+    if expected_runtime_scene is not None or expected_runtime_carrier is not None:
+        require_runtime_state_payload(
+            failures,
+            category=category,
+            context=context,
+            payload=payload.get("runtime_state"),
+            expected_scene=expected_runtime_scene,
+            expected_carrier=expected_runtime_carrier,
+            allowed_results={"pass", "block"},
+        )
+
+
+def check_skill_manifests(root: Path) -> list[Failure]:
+    failures: list[Failure] = []
+    expected_entries = {
+        "loom-init": "bootstrap/root",
+        "loom-adopt": "scenario/adopt",
+        "loom-resume": "scenario/resume",
+        "loom-pre-review": "scenario/pre-review",
+        "loom-review": "scenario/review",
+        "loom-handoff": "scenario/handoff",
+        "loom-retire": "scenario/retire",
+        "loom-merge-ready": "scenario/merge-ready",
+    }
+    registry_path = root / "skills/registry.json"
+    upgrade_contract_path = root / "skills/upgrade-contract.json"
+
+    for candidate in (registry_path, upgrade_contract_path):
+        if not candidate.exists():
+            return failures
+
+    try:
+        registry = load_json_file(registry_path)
+    except json.JSONDecodeError as exc:
+        return [Failure("skill-manifests", f"`skills/registry.json` is invalid JSON: {exc.msg}")]
+
+    try:
+        upgrade_contract = load_json_file(upgrade_contract_path)
+    except json.JSONDecodeError as exc:
+        return [Failure("skill-manifests", f"`skills/upgrade-contract.json` is invalid JSON: {exc.msg}")]
+
+    if not isinstance(registry, dict):
+        failures.append(Failure("skill-manifests", "`skills/registry.json` must be a JSON object"))
+        return failures
+    if not isinstance(upgrade_contract, dict):
+        failures.append(Failure("skill-manifests", "`skills/upgrade-contract.json` must be a JSON object"))
+        return failures
+
+    registry_version = registry.get("registry_version")
+    root_entry = registry.get("root_entry")
+    entries = registry.get("entries")
+    upgrade_reference = registry.get("upgrade_contract")
+    install_layout_reference = registry.get("install_layout")
+    layout_manifest: dict[str, object] | None = None
+    if registry_version != upgrade_contract.get("registry_version"):
+        failures.append(Failure("skill-manifests", "`skills/upgrade-contract.json` registry version must match `skills/registry.json`"))
+    if install_layout_reference != "install-layout.json":
+        failures.append(Failure("skill-manifests", "`skills/registry.json` must point `install_layout` to `install-layout.json`"))
+    else:
+        install_layout_path = registry_path.parent / install_layout_reference
+        if not install_layout_path.exists():
+            failures.append(Failure("skill-manifests", "`skills/install-layout.json` must exist"))
+        else:
+            try:
+                candidate_layout = load_json_file(install_layout_path)
+            except json.JSONDecodeError as exc:
+                failures.append(Failure("skill-manifests", f"`skills/install-layout.json` is invalid JSON: {exc.msg}"))
+            else:
+                layout_manifest = candidate_layout
+                required_paths = candidate_layout.get("required_paths")
+                if not isinstance(required_paths, list) or not required_paths:
+                    failures.append(Failure("skill-manifests", "`skills/install-layout.json` must declare a non-empty `required_paths`"))
+                else:
+                    for relative in required_paths:
+                        if not isinstance(relative, str) or not relative:
+                            failures.append(Failure("skill-manifests", "`skills/install-layout.json` required paths must be non-empty strings"))
+                            continue
+                        if not (registry_path.parent / relative).exists():
+                            failures.append(Failure("skill-manifests", f"`skills/install-layout.json` points to missing path `{relative}`"))
+                runtime_state = candidate_layout.get("runtime_state")
+                if not isinstance(runtime_state, dict):
+                    failures.append(Failure("skill-manifests", "`skills/install-layout.json` must declare `runtime_state`"))
+                else:
+                    recognized_states = runtime_state.get("recognized_states")
+                    if recognized_states != ["installed-runtime", "repo-local-demo", "upgrade-rehearsal"]:
+                        failures.append(
+                            Failure(
+                                "skill-manifests",
+                                "`skills/install-layout.json` runtime_state recognized_states must stay in the stable order",
+                            )
+                        )
+    if not isinstance(root_entry, str) or not root_entry:
+        failures.append(Failure("skill-manifests", "`skills/registry.json` must declare a non-empty `root_entry`"))
+        return failures
+    if not isinstance(entries, list) or not entries:
+        failures.append(Failure("skill-manifests", "`skills/registry.json` must declare at least one entry"))
+        return failures
+    if root_entry != "loom-init":
+        failures.append(Failure("skill-manifests", "`skills/registry.json` root entry must remain `loom-init`"))
+
+    root_registry_entry: dict[str, object] | None = None
+    seen_ids: set[str] = set()
+    for entry in entries:
+        if not isinstance(entry, dict):
+            failures.append(Failure("skill-manifests", "every registry entry must be an object"))
+            continue
+        entry_id = entry.get("id")
+        if not isinstance(entry_id, str) or not entry_id:
+            failures.append(Failure("skill-manifests", "every registry entry must declare a non-empty `id`"))
+            continue
+        if entry_id in seen_ids:
+            failures.append(Failure("skill-manifests", f"registry declares duplicate entry `{entry_id}`"))
+            continue
+        seen_ids.add(entry_id)
+        if entry_id == root_entry:
+            root_registry_entry = entry
+        expected_role = expected_entries.get(entry_id)
+        if expected_role is None:
+            failures.append(Failure("skill-manifests", f"registry declares unexpected entry `{entry_id}`"))
+        elif entry.get("role") != expected_role:
+            failures.append(Failure("skill-manifests", f"registry entry `{entry_id}` must declare role `{expected_role}`"))
+
+        for field in ("role", "contract_version", "manifest", "executable"):
+            value = entry.get(field)
+            if not isinstance(value, str) or not value:
+                failures.append(Failure("skill-manifests", f"registry entry `{entry_id}` must declare `{field}`"))
+
+        manifest_path = entry.get("manifest")
+        if not isinstance(manifest_path, str) or not manifest_path:
+            continue
+        manifest_file = registry_path.parent / manifest_path
+        if not manifest_file.exists():
+            failures.append(Failure("skill-manifests", f"registry entry `{entry_id}` points to missing manifest `{manifest_path}`"))
+            continue
+        executable_path = entry.get("executable")
+        if isinstance(executable_path, str) and executable_path:
+            if not (registry_path.parent / executable_path).resolve().exists():
+                failures.append(
+                    Failure("skill-manifests", f"registry entry `{entry_id}` points to missing executable `{executable_path}`")
+                )
+
+        try:
+            contract = load_json_file(manifest_file)
+        except json.JSONDecodeError as exc:
+            failures.append(Failure("skill-manifests", f"`{manifest_path}` is invalid JSON: {exc.msg}"))
+            continue
+        if not isinstance(contract, dict):
+            failures.append(Failure("skill-manifests", f"`{manifest_path}` must be a JSON object"))
+            continue
+
+        if contract.get("id") != entry_id:
+            failures.append(Failure("skill-manifests", f"`{manifest_path}` id must match registry entry `{entry_id}`"))
+        if contract.get("role") != entry.get("role"):
+            failures.append(Failure("skill-manifests", f"`{manifest_path}` role must match registry entry `{entry_id}`"))
+        if contract.get("contract_version") != entry.get("contract_version"):
+            failures.append(Failure("skill-manifests", f"`{manifest_path}` contract version must match registry entry `{entry_id}`"))
+
+        contract_root = contract.get("root_entry")
+        if entry_id == root_entry:
+            if contract_root is not True:
+                failures.append(Failure("skill-manifests", f"`{manifest_path}` must declare `root_entry: true`"))
+        elif contract_root is not False:
+            failures.append(Failure("skill-manifests", f"`{manifest_path}` must declare `root_entry: false`"))
+
+        entrypoint = contract.get("entrypoint")
+        if not isinstance(entrypoint, dict):
+            failures.append(Failure("skill-manifests", f"`{manifest_path}` must declare `entrypoint`"))
+        else:
+            required_entrypoint_keys = {"skill_markdown", "adapter_metadata"}
+            if entry_id == "loom-init":
+                required_entrypoint_keys.add("bootstrap_cli")
+                required_entrypoint_keys.add("route_cli")
+            else:
+                required_entrypoint_keys.add("orchestration_cli")
+            for key in required_entrypoint_keys:
+                value = entrypoint.get(key)
+                if not isinstance(value, str) or not value:
+                    failures.append(Failure("skill-manifests", f"`{manifest_path}` missing `entrypoint.{key}`"))
+                    continue
+                if not (manifest_file.parent / value).exists():
+                    failures.append(Failure("skill-manifests", f"`{manifest_path}` points `entrypoint.{key}` to missing `{value}`"))
+
+        for section in ("input_contract", "output_contract", "routing"):
+            value = contract.get(section)
+            if not isinstance(value, dict):
+                failures.append(Failure("skill-manifests", f"`{manifest_path}` must declare `{section}`"))
+                continue
+            reference = value.get("reference")
+            if not isinstance(reference, str) or not reference:
+                failures.append(Failure("skill-manifests", f"`{manifest_path}` must declare `{section}.reference`"))
+                continue
+            if not (manifest_file.parent / reference).exists():
+                failures.append(Failure("skill-manifests", f"`{manifest_path}` points `{section}.reference` to missing `{reference}`"))
+
+        output_contract = contract.get("output_contract")
+        if isinstance(output_contract, dict) and entry_id in GOVERNANCE_SURFACE_CONTRACT_SKILLS:
+            required_sections = output_contract.get("required_sections")
+            if not isinstance(required_sections, list):
+                failures.append(Failure("skill-manifests", f"`{manifest_path}` must declare `output_contract.required_sections`"))
+            elif "governance_surface" not in required_sections:
+                failures.append(
+                    Failure(
+                        "skill-manifests",
+                        f"`{manifest_path}` must require `governance_surface` in `output_contract.required_sections`",
+                    )
+                )
+
+        installation = contract.get("installation")
+        if not isinstance(installation, dict):
+            failures.append(Failure("skill-manifests", f"`{manifest_path}` must declare `installation`"))
+        else:
+            for field in ("registry", "upgrade_contract", "layout_manifest"):
+                value = installation.get(field)
+                if not isinstance(value, str) or not value:
+                    failures.append(Failure("skill-manifests", f"`{manifest_path}` must declare `installation.{field}`"))
+                    continue
+                if not (manifest_file.parent / value).exists():
+                    failures.append(Failure("skill-manifests", f"`{manifest_path}` points `installation.{field}` to missing `{value}`"))
+
+    if root_registry_entry is None:
+        failures.append(Failure("skill-manifests", f"`skills/registry.json` root entry `{root_entry}` does not match any declared entry"))
+        return failures
+    if seen_ids != set(expected_entries):
+        missing = sorted(set(expected_entries) - seen_ids)
+        extra = sorted(seen_ids - set(expected_entries))
+        if missing:
+            failures.append(Failure("skill-manifests", f"registry is missing first-wave entries: {', '.join(missing)}"))
+        if extra:
+            failures.append(Failure("skill-manifests", f"registry contains unexpected first-wave entries: {', '.join(extra)}"))
+    if upgrade_reference != "upgrade-contract.json":
+        failures.append(Failure("skill-manifests", "`skills/registry.json` must point to `upgrade-contract.json`"))
+
+    upgrade_root = upgrade_contract.get("root_entry")
+    current_contract_version = upgrade_contract.get("current_contract_version")
+    upgrade_policy = upgrade_contract.get("upgrade_policy")
+    if upgrade_root != root_entry:
+        failures.append(
+            Failure(
+                "skill-manifests",
+                f"`skills/upgrade-contract.json` root entry `{upgrade_root}` does not match registry root `{root_entry}`",
+            )
+        )
+    if current_contract_version != root_registry_entry.get("contract_version"):
+        failures.append(
+            Failure(
+                "skill-manifests",
+                "`skills/upgrade-contract.json` current contract version must match the registry entry version",
+            )
+        )
+        if not isinstance(upgrade_policy, dict):
+            failures.append(Failure("skill-manifests", "`skills/upgrade-contract.json` must declare `upgrade_policy`"))
+    else:
+        if upgrade_policy.get("mode") != "explicit":
+            failures.append(Failure("skill-manifests", "`upgrade_policy.mode` must be `explicit`"))
+        refresh_required = upgrade_policy.get("refresh_required")
+        if not isinstance(refresh_required, list) or not refresh_required:
+            failures.append(Failure("skill-manifests", "`upgrade_policy.refresh_required` must be a non-empty list"))
+        else:
+            required = {"registry", "manifest", "executable", "referenced_resources", "layout_manifest"}
+            if not required.issubset(set(refresh_required)):
+                failures.append(
+                    Failure(
+                        "skill-manifests",
+                        "`upgrade_policy.refresh_required` must cover registry, manifest, executable, referenced_resources, and layout_manifest",
+                    )
+                )
+
+    return failures
+
+
+def check_skill_routing(root: Path) -> list[Failure]:
+    failures: list[Failure] = []
+    target = root / "examples/new-project"
+    tool_path = root / "tools/loom_init.py"
+    if not tool_path.exists() or not target.exists():
+        return failures
+
+    registry = load_json_file(root / "skills/registry.json")
+    if not isinstance(registry, dict):
+        return failures
+    entries = registry.get("entries")
+    if not isinstance(entries, list):
+        return failures
+    explicit_skills = [
+        entry.get("id")
+        for entry in entries
+        if isinstance(entry, dict) and isinstance(entry.get("id"), str) and entry.get("id")
+    ]
+    for skill_id in explicit_skills:
+        payload, error = load_command_json(
+            root,
+            ["python3", "tools/loom_init.py", "route", "--target", "examples/new-project", "--skill", skill_id],
+        )
+        if error:
+            failures.append(Failure("skill-routing", f"explicit route for `{skill_id}` failed: {error}"))
+            continue
+        if payload.get("command") != "route":
+            failures.append(Failure("skill-routing", f"explicit route for `{skill_id}` must report `command: route`"))
+        if payload.get("result") != "pass":
+            failures.append(Failure("skill-routing", f"explicit route for `{skill_id}` must pass"))
+        if payload.get("selected_skill") != skill_id:
+            failures.append(Failure("skill-routing", f"explicit route for `{skill_id}` selected `{payload.get('selected_skill')}`"))
+        if payload.get("mode") != "explicit":
+            failures.append(Failure("skill-routing", f"explicit route for `{skill_id}` must report `mode: explicit`"))
+        if not isinstance(payload.get("summary"), str) or not payload.get("summary"):
+            failures.append(Failure("skill-routing", f"explicit route for `{skill_id}` must include `summary`"))
+        if not isinstance(payload.get("matched_signals"), list):
+            failures.append(Failure("skill-routing", f"explicit route for `{skill_id}` must include `matched_signals`"))
+        if not isinstance(payload.get("missing_inputs"), list):
+            failures.append(Failure("skill-routing", f"explicit route for `{skill_id}` must include `missing_inputs`"))
+        if payload.get("fallback_to") != "loom-init":
+            failures.append(Failure("skill-routing", f"explicit route for `{skill_id}` must keep `fallback_to: loom-init`"))
+        if skill_id in GOVERNANCE_SURFACE_ROUTE_SKILLS and payload.get("result") == "pass":
+            require_governance_surface(
+                failures,
+                category="skill-routing",
+                context=f"explicit route for `{skill_id}`",
+                payload=payload,
+            )
+
+    implicit_cases = (
+        ("请初始化这个新项目并接入 Loom", "loom-adopt"),
+        ("请接手当前事项并恢复上下文后继续推进", "loom-resume"),
+        ("请在进入 review 前做统一检查", "loom-pre-review"),
+        ("请对当前事项做正式 review 并给出审查结论", "loom-review"),
+        ("请准备交接并回写停点", "loom-handoff"),
+        ("请清理并 retire 当前事项现场", "loom-retire"),
+        ("请确认这个事项是否 merge-ready", "loom-merge-ready"),
+    )
+    for task, skill_id in implicit_cases:
+        payload, error = load_command_json(
+            root,
+            ["python3", "tools/loom_init.py", "route", "--target", "examples/new-project", "--task", task],
+        )
+        if error:
+            failures.append(Failure("skill-routing", f"implicit route for `{skill_id}` failed: {error}"))
+            continue
+        if payload.get("command") != "route":
+            failures.append(Failure("skill-routing", f"implicit route for `{skill_id}` must report `command: route`"))
+        if payload.get("result") != "pass":
+            failures.append(Failure("skill-routing", f"implicit route for `{skill_id}` must pass"))
+        if payload.get("selected_skill") != skill_id:
+            failures.append(Failure("skill-routing", f"implicit route for `{skill_id}` selected `{payload.get('selected_skill')}`"))
+        if payload.get("mode") != "implicit":
+            failures.append(Failure("skill-routing", f"implicit route for `{skill_id}` must report `mode: implicit`"))
+        if not isinstance(payload.get("matched_signals"), list) or not payload.get("matched_signals"):
+            failures.append(Failure("skill-routing", f"implicit route for `{skill_id}` must include matched signals"))
+        if not isinstance(payload.get("summary"), str) or not payload.get("summary"):
+            failures.append(Failure("skill-routing", f"implicit route for `{skill_id}` must include `summary`"))
+        if payload.get("fallback_to") != "loom-init":
+            failures.append(Failure("skill-routing", f"implicit route for `{skill_id}` must keep `fallback_to: loom-init`"))
+        if skill_id in GOVERNANCE_SURFACE_ROUTE_SKILLS and payload.get("result") == "pass":
+            require_governance_surface(
+                failures,
+                category="skill-routing",
+                context=f"implicit route for `{skill_id}`",
+                payload=payload,
+            )
+
+    fallback_payload, error = load_command_json(
+        root,
+        ["python3", "tools/loom_init.py", "route", "--target", "examples/new-project", "--task", "请帮我看看这个仓库"],
+    )
+    if error:
+        failures.append(Failure("skill-routing", f"fallback route failed: {error}"))
+    else:
+        if fallback_payload.get("result") != "fallback":
+            failures.append(Failure("skill-routing", "ambiguous task must return `fallback`"))
+        if fallback_payload.get("selected_skill") != "loom-init":
+            failures.append(Failure("skill-routing", "fallback route must select `loom-init`"))
+        if not isinstance(fallback_payload.get("missing_inputs"), list) or not fallback_payload.get("missing_inputs"):
+            failures.append(Failure("skill-routing", "fallback route must include `missing_inputs`"))
+
+    ambiguous_payload, error = load_command_json(
+        root,
+        [
+            "python3",
+            "tools/loom_init.py",
+            "route",
+            "--target",
+            "examples/new-project",
+            "--task",
+            "请接手当前事项并在 review 前检查",
+        ],
+    )
+    if error:
+        failures.append(Failure("skill-routing", f"ambiguous route failed: {error}"))
+    else:
+        if ambiguous_payload.get("result") != "fallback":
+            failures.append(Failure("skill-routing", "multi-match task must return `fallback`"))
+        if ambiguous_payload.get("selected_skill") != "loom-init":
+            failures.append(Failure("skill-routing", "multi-match route must select `loom-init`"))
+        if not isinstance(ambiguous_payload.get("matched_signals"), list) or len(ambiguous_payload.get("matched_signals", [])) < 2:
+            failures.append(Failure("skill-routing", "multi-match route must expose matched signals"))
+
+    unknown_payload, error = load_command_json(
+        root,
+        ["python3", "tools/loom_init.py", "route", "--target", "examples/new-project", "--skill", "not-a-skill"],
+    )
+    if error:
+        failures.append(Failure("skill-routing", f"unknown explicit route failed: {error}"))
+    else:
+        if unknown_payload.get("result") != "block":
+            failures.append(Failure("skill-routing", "unknown explicit skill must block"))
+        if unknown_payload.get("selected_skill") != "loom-init":
+            failures.append(Failure("skill-routing", "unknown explicit skill must fall back to `loom-init`"))
+
+    return failures
+
+
+def check_demo_assets(root: Path) -> list[Failure]:
+    failures = check_required_paths(root, "demo-assets", DEMO_ASSETS)
+
+    init_result_path = root / "examples/new-project/.loom/bootstrap/init-result.json"
+    if init_result_path.exists():
+        try:
+            init_result = load_json_file(init_result_path)
+        except json.JSONDecodeError as exc:
+            failures.append(Failure("demo-assets", f"demo init-result is invalid JSON: {exc.msg}"))
+            return failures
+        if not isinstance(init_result, dict):
+            failures.append(Failure("demo-assets", "demo init-result must be a JSON object"))
+            return failures
+        run = init_result.get("run")
+        if not isinstance(run, dict) or run.get("scenario_key") != "new":
+            failures.append(Failure("demo-assets", "demo init-result must keep `scenario_key` as `new`"))
+    return failures
+
+
+def check_demo_fact_chain(root: Path) -> list[Failure]:
+    target = root / "examples/new-project"
+    if not target.exists():
+        return []
+
+    report, errors = inspect_fact_chain(target)
+    failures: list[Failure] = []
+    for detail in errors:
+        failures.append(Failure("demo-fact-chain", detail))
+    if report and report.get("fact_chain", {}).get("entry_points", {}).get("status_surface") != ".loom/status/current.md":
+        failures.append(Failure("demo-fact-chain", "demo fact chain must point status_surface to `.loom/status/current.md`"))
+    return failures
+
+
+def check_demo_repo_local_cli(root: Path) -> list[Failure]:
+    failures: list[Failure] = []
+    target = root / "examples/new-project"
+    if not target.exists():
+        return failures
+
+    repo_local_commands = [
+        (
+            "repo-local-verify",
+            ["python3", ".loom/bin/loom_init.py", "verify", "--target", "."],
+            "ok",
+        ),
+        (
+            "repo-local-fact-chain",
+            ["python3", ".loom/bin/loom_init.py", "fact-chain", "--target", "."],
+            "ok",
+        ),
+    ]
+    for label, args, expected_key in repo_local_commands:
+        payload, error = load_command_json(root, args, cwd=target)
+        if error:
+            failures.append(Failure("demo-repo-local-cli", f"`{label}` failed: {error}"))
+            continue
+        if payload.get(expected_key) is not True:
+            failures.append(Failure("demo-repo-local-cli", f"`{label}` must report `{expected_key}: true`"))
+    return failures
+
+
+def check_deep_existing_repo_bootstrap(root: Path) -> list[Failure]:
+    failures: list[Failure] = []
+    with tempfile.TemporaryDirectory(prefix="loom-check-deep-existing-") as tmp:
+        tmp_root = Path(tmp)
+
+        def write_repo(target: Path, *, validation_entry: bool, pr_template: bool, workflow_doc: bool) -> None:
+            (target / ".github" / "workflows").mkdir(parents=True, exist_ok=True)
+            (target / "scripts").mkdir(parents=True, exist_ok=True)
+            (target / "src").mkdir(parents=True, exist_ok=True)
+            (target / "README.md").write_text("# Sample Repo\n", encoding="utf-8")
+            (target / "AGENTS.md").write_text("# Root Rules\n", encoding="utf-8")
+            (target / "src" / "main.py").write_text("print('ok')\n", encoding="utf-8")
+            (target / "scripts" / "governance_status.py").write_text("print('ok')\n", encoding="utf-8")
+            (target / ".github" / "workflows" / "ci.yml").write_text("name: ci\n", encoding="utf-8")
+            if workflow_doc:
+                (target / "WORKFLOW.md").write_text("# Workflow\n", encoding="utf-8")
+            if validation_entry:
+                (target / "Makefile").write_text("check:\n\t@echo ok\n", encoding="utf-8")
+            if pr_template:
+                (target / ".github" / "PULL_REQUEST_TEMPLATE.md").write_text("## Summary\n", encoding="utf-8")
+
+        deep_target = tmp_root / "deep-existing"
+        write_repo(deep_target, validation_entry=True, pr_template=True, workflow_doc=True)
+        deep_payload, deep_error = load_command_json(
+            root,
+            [
+                "python3",
+                "tools/loom_init.py",
+                "bootstrap",
+                "--target",
+                str(deep_target),
+                "--write",
+                "--force",
+                "--verify",
+                "--install-pr-template",
+            ],
+        )
+        if deep_error:
+            failures.append(Failure("deep-existing-bootstrap", f"`deep-existing bootstrap` failed: {deep_error}"))
+        else:
+            recommended = deep_payload.get("recommended_adoption")
+            verification = deep_payload.get("verification")
+            governance_surface = deep_payload.get("governance_surface")
+            if not isinstance(recommended, dict) or recommended.get("path") != "deep-existing-repo":
+                failures.append(Failure("deep-existing-bootstrap", "`deep-existing bootstrap` must select `recommended_adoption.path = deep-existing-repo`"))
+            run = deep_payload.get("run")
+            if not isinstance(run, dict) or run.get("scenario_key") != "complex-existing":
+                failures.append(Failure("deep-existing-bootstrap", "`deep-existing bootstrap` must keep `scenario_key = complex-existing`"))
+            if not isinstance(verification, dict) or verification.get("ok") is not True:
+                failures.append(Failure("deep-existing-bootstrap", "`deep-existing bootstrap` must verify successfully"))
+            if not isinstance(governance_surface, dict) or governance_surface.get("repository_mode") != "complex-existing":
+                failures.append(Failure("deep-existing-bootstrap", "`deep-existing bootstrap` must keep `governance_surface.repository_mode = complex-existing`"))
+            for required in (
+                ".loom/companion/README.md",
+                ".loom/companion/checkpoints.md",
+                ".loom/companion/review.md",
+                ".loom/companion/merge-ready.md",
+                ".loom/companion/closeout.md",
+            ):
+                if not (deep_target / required).exists():
+                    failures.append(Failure("deep-existing-bootstrap", f"`deep-existing bootstrap` is missing `{required}`"))
+            for forbidden in (
+                ".loom/work-items/INIT-0001.md",
+                ".loom/progress/INIT-0001.md",
+                ".loom/status/current.md",
+            ):
+                if (deep_target / forbidden).exists():
+                    failures.append(Failure("deep-existing-bootstrap", f"`deep-existing bootstrap` must not generate `{forbidden}`"))
+            fact_chain = deep_payload.get("fact_chain")
+            if not isinstance(fact_chain, dict) or fact_chain.get("mode") != "repo-native attach-only":
+                failures.append(Failure("deep-existing-bootstrap", "`deep-existing bootstrap` must keep `fact_chain.mode = repo-native attach-only`"))
+
+        full_target = tmp_root / "full-bootstrap"
+        write_repo(full_target, validation_entry=False, pr_template=False, workflow_doc=False)
+        full_payload, full_error = load_command_json(
+            root,
+            [
+                "python3",
+                "tools/loom_init.py",
+                "bootstrap",
+                "--target",
+                str(full_target),
+                "--write",
+                "--force",
+                "--verify",
+                "--install-pr-template",
+            ],
+        )
+        if full_error:
+            failures.append(Failure("deep-existing-bootstrap", f"`full-bootstrap fallback sample` failed: {full_error}"))
+        else:
+            recommended = full_payload.get("recommended_adoption")
+            if not isinstance(recommended, dict) or recommended.get("path") != "full-bootstrap":
+                failures.append(Failure("deep-existing-bootstrap", "complex existing sample without overload must keep `recommended_adoption.path = full-bootstrap`"))
+            for required in (
+                ".loom/work-items/INIT-0001.md",
+                ".loom/progress/INIT-0001.md",
+                ".loom/status/current.md",
+            ):
+                if not (full_target / required).exists():
+                    failures.append(Failure("deep-existing-bootstrap", f"`full-bootstrap fallback sample` must generate `{required}`"))
+    return failures
+
+
+def check_daily_execution_cli(root: Path) -> list[Failure]:
+    failures: list[Failure] = []
+    example_target = root / "examples/new-project"
+    tool_path = root / "tools/loom_flow.py"
+    if not tool_path.exists() or not example_target.exists():
+        return failures
+
+    demo_commands = [
+        (
+            "runtime-state-init",
+            ["python3", "tools/loom_init.py", "runtime-state", "--target", "."],
+            {"pass"},
+        ),
+        (
+            "runtime-state-flow",
+            ["python3", "tools/loom_flow.py", "runtime-state", "--target", "examples/new-project", "--item", "INIT-0001"],
+            {"pass"},
+        ),
+        (
+            "fact-chain",
+            ["python3", "tools/loom_flow.py", "fact-chain", "--target", "examples/new-project", "--item", "INIT-0001"],
+            {"pass"},
+        ),
+        (
+            "runtime-evidence",
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "runtime-evidence",
+                "--target",
+                "examples/new-project",
+                "--item",
+                "INIT-0001",
+            ],
+            {"pass"},
+        ),
+        (
+            "state-check",
+            ["python3", "tools/loom_flow.py", "state-check", "--target", "examples/new-project", "--item", "INIT-0001"],
+            {"pass"},
+        ),
+        (
+            "flow-pre-review",
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "flow",
+                "pre-review",
+                "--target",
+                "examples/new-project",
+                "--item",
+                "INIT-0001",
+            ],
+            {"pass", "block", "fallback"},
+        ),
+        (
+            "flow-review",
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "flow",
+                "review",
+                "--target",
+                "examples/new-project",
+                "--item",
+                "INIT-0001",
+            ],
+            {"pass", "block", "fallback"},
+        ),
+        (
+            "flow-resume",
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "flow",
+                "resume",
+                "--target",
+                "examples/new-project",
+                "--item",
+                "INIT-0001",
+            ],
+            {"pass"},
+        ),
+        (
+            "flow-handoff",
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "flow",
+                "handoff",
+                "--target",
+                "examples/new-project",
+                "--item",
+                "INIT-0001",
+            ],
+            {"pass", "block"},
+        ),
+        (
+            "flow-merge-ready",
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "flow",
+                "merge-ready",
+                "--target",
+                "examples/new-project",
+                "--item",
+                "INIT-0001",
+            ],
+            {"pass", "block", "fallback"},
+        ),
+        (
+            "admission",
+            ["python3", "tools/loom_flow.py", "checkpoint", "admission", "--target", "examples/new-project", "--item", "INIT-0001"],
+            {"pass"},
+        ),
+        (
+            "build",
+            ["python3", "tools/loom_flow.py", "checkpoint", "build", "--target", "examples/new-project", "--item", "INIT-0001"],
+            {"pass", "block", "fallback"},
+        ),
+        (
+            "merge",
+            ["python3", "tools/loom_flow.py", "checkpoint", "merge", "--target", "examples/new-project", "--item", "INIT-0001"],
+            {"pass", "block", "fallback"},
+        ),
+        (
+            "locate",
+            ["python3", "tools/loom_flow.py", "workspace", "locate", "--target", "examples/new-project", "--item", "INIT-0001"],
+            {"pass"},
+        ),
+        (
+            "review-read",
+            ["python3", "tools/loom_flow.py", "review", "read", "--target", "examples/new-project", "--item", "INIT-0001"],
+            {"pass"},
+        ),
+        (
+            "host-lifecycle",
+            ["python3", "tools/loom_flow.py", "host-lifecycle", "--target", "examples/new-project", "--item", "INIT-0001"],
+            {"pass"},
+        ),
+        (
+            "closeout-check",
+            ["python3", "tools/loom_flow.py", "closeout", "check", "--target", ".", "--skip-gate"],
+            {"pass"},
+        ),
+        (
+            "closeout-sync",
+            ["python3", "tools/loom_flow.py", "closeout", "sync", "--target", ".", "--skip-gate"],
+            {"pass"},
+        ),
+        (
+            "reconciliation-audit",
+            ["python3", "tools/loom_flow.py", "reconciliation", "audit", "--target", "."],
+            {"block"},
+        ),
+        (
+            "purity",
+            ["python3", "tools/loom_flow.py", "purity-check", "--target", "examples/new-project", "--item", "INIT-0001"],
+            {"pass"},
+        ),
+    ]
+    for label, args, allowed_results in demo_commands:
+        payload, error = load_command_json(root, args)
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`{label}` command failed: {error}"))
+            continue
+        result = payload.get("result")
+        if result not in allowed_results:
+            failures.append(
+                Failure(
+                    "daily-execution-cli",
+                    f"`{label}` returned unexpected result `{result}`",
+                )
+            )
+        if label == "runtime-state-init":
+            if payload.get("command") != "runtime-state":
+                failures.append(Failure("daily-execution-cli", "`loom-init runtime-state` must report `command: runtime-state`"))
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`loom-init runtime-state`",
+                payload=payload.get("runtime_state"),
+                expected_scene="repo-local-demo",
+                expected_carrier="repo-local-wrapper",
+                allowed_results={"pass"},
+            )
+        if label == "runtime-state-flow":
+            if payload.get("command") != "runtime-state":
+                failures.append(Failure("daily-execution-cli", "`loom-flow runtime-state` must report `command: runtime-state`"))
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`loom-flow runtime-state`",
+                payload=payload.get("runtime_state"),
+                expected_scene="repo-local-demo",
+                expected_carrier="repo-local-wrapper",
+                allowed_results={"pass"},
+            )
+        if label == "runtime-evidence":
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`runtime-evidence`",
+                payload=payload.get("runtime_state"),
+                expected_scene="repo-local-demo",
+                expected_carrier="repo-local-wrapper",
+                allowed_results={"pass"},
+            )
+        if label == "state-check":
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`state-check`",
+                payload=payload.get("runtime_state"),
+                expected_scene="repo-local-demo",
+                expected_carrier="repo-local-wrapper",
+                allowed_results={"pass"},
+            )
+        if label == "flow-pre-review":
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`flow pre-review`",
+                payload=payload.get("runtime_state"),
+                expected_scene="repo-local-demo",
+                expected_carrier="repo-local-wrapper",
+                allowed_results={"pass"},
+            )
+            steps = payload.get("steps")
+            if isinstance(steps, list):
+                step_names = [step.get("name") for step in steps if isinstance(step, dict)]
+                if step_names != [
+                    "runtime-state",
+                    "fact-chain",
+                    "state-check",
+                    "runtime-evidence",
+                    "checkpoint-admission",
+                    "workspace-locate",
+                ]:
+                    failures.append(
+                        Failure(
+                            "daily-execution-cli",
+                            "`flow pre-review` must run runtime-state, fact-chain, state-check, runtime-evidence, checkpoint-admission, and workspace-locate in order",
+                        )
+                    )
+        if label == "purity":
+            purity = payload.get("purity")
+            if not isinstance(purity, dict):
+                failures.append(Failure("daily-execution-cli", "`purity` output must include a `purity` object"))
+                continue
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`purity`",
+                payload=payload.get("runtime_state"),
+                expected_scene="repo-local-demo",
+                expected_carrier="repo-local-wrapper",
+                allowed_results={"pass"},
+            )
+            scope_assessment = purity.get("scope_assessment")
+            if not isinstance(scope_assessment, dict):
+                failures.append(Failure("daily-execution-cli", "`purity` output must include `scope_assessment`"))
+                continue
+            mode = scope_assessment.get("mode")
+            if mode not in {"constrained", "unconstrained"}:
+                failures.append(
+                    Failure("daily-execution-cli", "`scope_assessment.mode` must be `constrained` or `unconstrained`")
+                )
+        if label == "flow-resume":
+            if payload.get("command") != "flow":
+                failures.append(Failure("daily-execution-cli", "`flow resume` must report `command: flow`"))
+            if payload.get("operation") != "resume":
+                failures.append(Failure("daily-execution-cli", "`flow resume` must report `operation: resume`"))
+            if not isinstance(payload.get("summary"), str) or not payload.get("summary"):
+                failures.append(Failure("daily-execution-cli", "`flow resume` must include a non-empty `summary`"))
+            if not isinstance(payload.get("missing_inputs"), list):
+                failures.append(Failure("daily-execution-cli", "`flow resume` must include `missing_inputs`"))
+            if payload.get("fallback_to") not in {None, "admission"}:
+                failures.append(Failure("daily-execution-cli", "`flow resume` fallback must be `null` or `admission`"))
+            for key in ("item", "workspace", "recovery", "checkpoint", "state_check"):
+                if not isinstance(payload.get(key), dict):
+                    failures.append(Failure("daily-execution-cli", f"`flow resume` must include `{key}`"))
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`flow resume`",
+                payload=payload.get("runtime_state"),
+                expected_scene="repo-local-demo",
+                expected_carrier="repo-local-wrapper",
+                allowed_results={"pass"},
+            )
+            require_governance_surface(
+                failures,
+                category="daily-execution-cli",
+                context="`flow resume`",
+                payload=payload,
+            )
+            steps = payload.get("steps")
+            if not isinstance(steps, list):
+                failures.append(Failure("daily-execution-cli", "`flow resume` must include `steps`"))
+                continue
+            step_names = [step.get("name") for step in steps if isinstance(step, dict)]
+            if step_names != ["runtime-state", "fact-chain", "state-check", "workspace-locate"]:
+                failures.append(
+                    Failure(
+                        "daily-execution-cli",
+                        "`flow resume` must run runtime-state, fact-chain, state-check, and workspace-locate in order",
+                    )
+                )
+            recovery = payload.get("recovery")
+            if isinstance(recovery, dict):
+                for field in ("current_stop", "next_step", "blockers", "latest_validation_summary"):
+                    value = recovery.get(field)
+                    if not isinstance(value, str) or not value:
+                        failures.append(
+                            Failure("daily-execution-cli", f"`flow resume` recovery must include non-empty `{field}`")
+                        )
+            checkpoint = payload.get("checkpoint")
+            if isinstance(checkpoint, dict):
+                if checkpoint.get("normalized") not in {"admission", "build", "merge", "retired"}:
+                    failures.append(
+                        Failure("daily-execution-cli", "`flow resume` checkpoint must include a known normalized value")
+                    )
+            state_check = payload.get("state_check")
+            if isinstance(state_check, dict):
+                if state_check.get("result") not in {"pass", "block"}:
+                    failures.append(
+                        Failure("daily-execution-cli", "`flow resume` state_check.result must be `pass` or `block`")
+                    )
+                if not isinstance(state_check.get("checks"), dict):
+                    failures.append(Failure("daily-execution-cli", "`flow resume` must include `state_check.checks`"))
+        if label == "flow-handoff":
+            if payload.get("command") != "flow":
+                failures.append(Failure("daily-execution-cli", "`flow handoff` must report `command: flow`"))
+            if payload.get("operation") != "handoff":
+                failures.append(Failure("daily-execution-cli", "`flow handoff` must report `operation: handoff`"))
+            if not isinstance(payload.get("summary"), str) or not payload.get("summary"):
+                failures.append(Failure("daily-execution-cli", "`flow handoff` must include a non-empty `summary`"))
+            if not isinstance(payload.get("missing_inputs"), list):
+                failures.append(Failure("daily-execution-cli", "`flow handoff` must include `missing_inputs`"))
+            if payload.get("fallback_to") not in {None, "admission"}:
+                failures.append(Failure("daily-execution-cli", "`flow handoff` fallback must be `null` or `admission`"))
+            for key in ("item", "workspace", "checkpoint", "state_check"):
+                if not isinstance(payload.get(key), dict):
+                    failures.append(Failure("daily-execution-cli", f"`flow handoff` must include `{key}`"))
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`flow handoff`",
+                payload=payload.get("runtime_state"),
+                expected_scene="repo-local-demo",
+                expected_carrier="repo-local-wrapper",
+                allowed_results={"pass"},
+            )
+            for key in (
+                "recovery_entry",
+                "status_surface",
+                "current_stop",
+                "next_step",
+                "blockers",
+                "latest_validation_summary",
+            ):
+                value = payload.get(key)
+                if not isinstance(value, str) or not value:
+                    failures.append(Failure("daily-execution-cli", f"`flow handoff` must include non-empty `{key}`"))
+            if payload.get("fallback_target") not in {None, "admission"}:
+                failures.append(Failure("daily-execution-cli", "`flow handoff` fallback_target must be `null` or `admission`"))
+            writeback_fields = payload.get("writeback_fields")
+            if writeback_fields != [
+                "current_stop",
+                "next_step",
+                "blockers",
+                "latest_validation_summary",
+            ]:
+                failures.append(
+                    Failure(
+                        "daily-execution-cli",
+                        "`flow handoff` must report the stable writeback field list in order",
+                    )
+                )
+            steps = payload.get("steps")
+            if not isinstance(steps, list):
+                failures.append(Failure("daily-execution-cli", "`flow handoff` must include `steps`"))
+                continue
+            step_names = [step.get("name") for step in steps if isinstance(step, dict)]
+            if step_names != ["runtime-state", "fact-chain", "state-check", "workspace-locate"]:
+                failures.append(
+                    Failure(
+                        "daily-execution-cli",
+                        "`flow handoff` must run runtime-state, fact-chain, state-check, and workspace-locate in order",
+                    )
+                )
+            state_check = payload.get("state_check")
+            if isinstance(state_check, dict):
+                if state_check.get("result") not in {"pass", "block"}:
+                    failures.append(
+                        Failure("daily-execution-cli", "`flow handoff` state_check.result must be `pass` or `block`")
+                    )
+                if not isinstance(state_check.get("checks"), dict):
+                    failures.append(Failure("daily-execution-cli", "`flow handoff` must include `state_check.checks`"))
+        if label == "flow-review":
+            if payload.get("command") != "flow":
+                failures.append(Failure("daily-execution-cli", "`flow review` must report `command: flow`"))
+            if payload.get("operation") != "review":
+                failures.append(Failure("daily-execution-cli", "`flow review` must report `operation: review`"))
+            for key in ("item", "state_check", "runtime_evidence", "build_checkpoint", "review", "current_checkpoint", "repo_specific_requirements"):
+                if not isinstance(payload.get(key), dict):
+                    failures.append(Failure("daily-execution-cli", f"`flow review` must include `{key}`"))
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`flow review`",
+                payload=payload.get("runtime_state"),
+                expected_scene="repo-local-demo",
+                expected_carrier="repo-local-wrapper",
+                allowed_results={"pass"},
+            )
+            steps = payload.get("steps")
+            if not isinstance(steps, list):
+                failures.append(Failure("daily-execution-cli", "`flow review` must include `steps`"))
+                continue
+            step_names = [step.get("name") for step in steps if isinstance(step, dict)]
+            if step_names != [
+                "runtime-state",
+                "fact-chain",
+                "state-check",
+                "runtime-evidence",
+                "checkpoint-build",
+                "review-entry",
+            ]:
+                failures.append(
+                    Failure(
+                        "daily-execution-cli",
+                        "`flow review` must run runtime-state, fact-chain, state-check, runtime-evidence, checkpoint-build, and review-entry in order",
+                    )
+                )
+            review = payload.get("review")
+            if isinstance(review, dict):
+                require_review_record_contract(
+                    failures,
+                    category="daily-execution-cli",
+                    context="`flow review` review.record",
+                    payload=review.get("record"),
+                )
+            require_repo_specific_requirements_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`flow review` repo_specific_requirements",
+                payload=payload.get("repo_specific_requirements"),
+                expected_surface="review",
+            )
+        if label == "review-read":
+            if payload.get("command") != "review":
+                failures.append(Failure("daily-execution-cli", "`review read` must report `command: review`"))
+            if payload.get("operation") != "read":
+                failures.append(Failure("daily-execution-cli", "`review read` must report `operation: read`"))
+            review = payload.get("review")
+            if not isinstance(review, dict):
+                failures.append(Failure("daily-execution-cli", "`review read` must include a `review` object"))
+            elif not isinstance(review.get("record"), dict):
+                failures.append(Failure("daily-execution-cli", "`review read` must include `review.record`"))
+            else:
+                require_review_record_contract(
+                    failures,
+                    category="daily-execution-cli",
+                    context="`review read` review.record",
+                    payload=review.get("record"),
+                )
+        if label == "host-lifecycle":
+            if payload.get("command") != "host-lifecycle":
+                failures.append(Failure("daily-execution-cli", "`host-lifecycle` must report `command: host-lifecycle`"))
+            require_host_lifecycle_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`host-lifecycle`",
+                payload=payload,
+            )
+        if label in {"closeout-check", "closeout-sync"}:
+            if payload.get("command") != "closeout":
+                failures.append(Failure("daily-execution-cli", f"`{label}` must report `command: closeout`"))
+            expected_operation = "check" if label == "closeout-check" else "sync"
+            if payload.get("operation") != expected_operation:
+                failures.append(
+                    Failure("daily-execution-cli", f"`{label}` must report `operation: {expected_operation}`")
+                )
+            repo = payload.get("repo")
+            if not isinstance(repo, dict):
+                failures.append(Failure("daily-execution-cli", f"`{label}` must include `repo`"))
+            else:
+                if not isinstance(repo.get("owner"), str) or not repo.get("owner"):
+                    failures.append(Failure("daily-execution-cli", f"`{label}` must include `repo.owner`"))
+                if not isinstance(repo.get("name"), str) or not repo.get("name"):
+                    failures.append(Failure("daily-execution-cli", f"`{label}` must include `repo.name`"))
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context=f"`{label}`",
+                payload=payload.get("runtime_state"),
+                expected_scene="repo-local-demo",
+                expected_carrier="repo-local-wrapper",
+                allowed_results={"pass"},
+            )
+            require_closeout_reconciliation_contract(
+                failures,
+                category="daily-execution-cli",
+                context=f"`{label}`",
+                payload=payload,
+            )
+            require_repo_specific_requirements_payload(
+                failures,
+                category="daily-execution-cli",
+                context=f"`{label}` repo_specific_requirements",
+                payload=payload.get("repo_specific_requirements"),
+                expected_surface="closeout",
+            )
+        if label == "reconciliation-audit":
+            if payload.get("command") != "reconciliation":
+                failures.append(Failure("daily-execution-cli", "`reconciliation audit` must report `command: reconciliation`"))
+            if payload.get("operation") != "audit":
+                failures.append(Failure("daily-execution-cli", "`reconciliation audit` must report `operation: audit`"))
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`reconciliation audit`",
+                payload=payload.get("runtime_state"),
+                expected_scene="repo-local-demo",
+                expected_carrier="repo-local-wrapper",
+                allowed_results={"pass"},
+            )
+            require_reconciliation_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`reconciliation audit`",
+                payload=payload,
+            )
+        if label == "flow-merge-ready":
+            if payload.get("command") != "flow":
+                failures.append(Failure("daily-execution-cli", "`flow merge-ready` must report `command: flow`"))
+            if payload.get("operation") != "merge-ready":
+                failures.append(Failure("daily-execution-cli", "`flow merge-ready` must report `operation: merge-ready`"))
+            if not isinstance(payload.get("summary"), str) or not payload.get("summary"):
+                failures.append(Failure("daily-execution-cli", "`flow merge-ready` must include a non-empty `summary`"))
+            if not isinstance(payload.get("missing_inputs"), list):
+                failures.append(Failure("daily-execution-cli", "`flow merge-ready` must include `missing_inputs`"))
+            if payload.get("fallback_to") not in {None, "admission", "build", "merge", "retired"}:
+                failures.append(
+                    Failure(
+                        "daily-execution-cli",
+                        "`flow merge-ready` fallback must be `null` or a known checkpoint",
+                    )
+                )
+            for key in ("item", "runtime_state", "state_check", "runtime_evidence", "build_checkpoint", "merge_checkpoint", "current_checkpoint", "repo_specific_requirements"):
+                if not isinstance(payload.get(key), dict):
+                    failures.append(Failure("daily-execution-cli", f"`flow merge-ready` must include `{key}`"))
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`flow merge-ready`",
+                payload=payload.get("runtime_state"),
+                expected_scene="repo-local-demo",
+                expected_carrier="repo-local-wrapper",
+                allowed_results={"pass"},
+            )
+            if not isinstance(payload.get("current_lane"), str) or not payload.get("current_lane"):
+                failures.append(Failure("daily-execution-cli", "`flow merge-ready` must include `current_lane`"))
+            if not isinstance(payload.get("latest_validation_summary"), str) or not payload.get("latest_validation_summary"):
+                failures.append(
+                    Failure("daily-execution-cli", "`flow merge-ready` must include `latest_validation_summary`")
+                )
+            steps = payload.get("steps")
+            if not isinstance(steps, list):
+                failures.append(Failure("daily-execution-cli", "`flow merge-ready` must include `steps`"))
+                continue
+            step_names = [step.get("name") for step in steps if isinstance(step, dict)]
+            if step_names != [
+                "runtime-state",
+                "fact-chain",
+                "state-check",
+                "runtime-evidence",
+                "checkpoint-build",
+                "checkpoint-merge",
+            ]:
+                failures.append(
+                    Failure(
+                        "daily-execution-cli",
+                        "`flow merge-ready` must run runtime-state, fact-chain, state-check, runtime-evidence, checkpoint-build, and checkpoint-merge in order",
+                    )
+                )
+            state_check = payload.get("state_check")
+            if isinstance(state_check, dict):
+                if state_check.get("result") not in {"pass", "block"}:
+                    failures.append(
+                        Failure("daily-execution-cli", "`flow merge-ready` state_check.result must be `pass` or `block`")
+                    )
+                if not isinstance(state_check.get("checks"), dict):
+                    failures.append(Failure("daily-execution-cli", "`flow merge-ready` must include `state_check.checks`"))
+            runtime_evidence = payload.get("runtime_evidence")
+            if isinstance(runtime_evidence, dict):
+                for field in ("run_entry", "logs_entry", "diagnostics_entry", "verification_entry", "lane_entry"):
+                    if not isinstance(runtime_evidence.get(field), dict):
+                        failures.append(
+                            Failure("daily-execution-cli", f"`flow merge-ready` must include runtime evidence field `{field}`")
+                        )
+            for key in ("build_checkpoint", "merge_checkpoint"):
+                checkpoint = payload.get(key)
+                if isinstance(checkpoint, dict):
+                    if checkpoint.get("result") not in {"pass", "block", "fallback"}:
+                        failures.append(
+                            Failure(
+                                "daily-execution-cli",
+                                f"`flow merge-ready` {key}.result must be `pass`, `block`, or `fallback`",
+                            )
+                        )
+                    if not isinstance(checkpoint.get("missing_inputs"), list):
+                        failures.append(
+                            Failure("daily-execution-cli", f"`flow merge-ready` {key} must include `missing_inputs`")
+                        )
+            merge_checkpoint = payload.get("merge_checkpoint")
+            if isinstance(merge_checkpoint, dict) and not isinstance(merge_checkpoint.get("pr_template"), dict):
+                failures.append(Failure("daily-execution-cli", "`flow merge-ready` must include `merge_checkpoint.pr_template`"))
+            require_repo_specific_requirements_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`flow merge-ready` repo_specific_requirements",
+                payload=payload.get("repo_specific_requirements"),
+                expected_surface="merge_ready",
+            )
+            if payload.get("result") != "fallback":
+                failures.append(Failure("daily-execution-cli", "`flow merge-ready` must return `fallback` for the bootstrap demo"))
+            if payload.get("fallback_to") != "admission":
+                failures.append(Failure("daily-execution-cli", "`flow merge-ready` must fall back to `admission` for the bootstrap demo"))
+            if isinstance(payload.get("build_checkpoint"), dict) and payload["build_checkpoint"].get("fallback_to") != "admission":
+                failures.append(Failure("daily-execution-cli", "`flow merge-ready` build checkpoint must fall back to `admission` for the bootstrap demo"))
+            if isinstance(payload.get("merge_checkpoint"), dict) and payload["merge_checkpoint"].get("fallback_to") != "admission":
+                failures.append(Failure("daily-execution-cli", "`flow merge-ready` merge checkpoint must fall back to `admission` for the bootstrap demo"))
+
+    with tempfile.TemporaryDirectory(prefix="loom-check-review-run-") as tmp:
+        source_snapshot = Path(tmp) / "source-snapshot"
+        review_target = Path(tmp) / "new-project"
+        fake_bin = Path(tmp) / "bin"
+        fake_bin.mkdir(parents=True, exist_ok=True)
+        shutil.copytree(root, source_snapshot, ignore=shutil.ignore_patterns(".git", ".DS_Store", "__pycache__"))
+
+        def prepare_review_target(target: Path, label: str) -> bool:
+            shutil.copytree(source_snapshot, target)
+            for args in (
+                ["git", "init"],
+                ["git", "config", "user.email", "loom-check@example.com"],
+                ["git", "config", "user.name", "loom-check"],
+            ):
+                result = run_command(root, args, cwd=target)
+                if result.returncode != 0:
+                    detail = result.stderr.strip() or result.stdout.strip() or "git setup failed"
+                    failures.append(Failure("daily-execution-cli", f"`{label}` setup failed: {detail}"))
+                    return False
+            payload, error = load_command_json(
+                root,
+                [
+                    "python3",
+                    "tools/loom_init.py",
+                    "bootstrap",
+                    "--target",
+                    ".",
+                    "--write",
+                    "--force",
+                    "--verify",
+                    "--install-pr-template",
+                ],
+                cwd=target,
+            )
+            if error:
+                failures.append(Failure("daily-execution-cli", f"`{label}` bootstrap failed: {error}"))
+                return False
+            verification = payload.get("verification")
+            if not isinstance(verification, dict) or verification.get("ok") is not True:
+                failures.append(Failure("daily-execution-cli", f"`{label}` bootstrap must verify successfully"))
+                return False
+            for args in (
+                ["git", "add", "."],
+                ["git", "add", "-f", ".loom"],
+                ["git", "commit", "-m", "review-run baseline"],
+            ):
+                result = run_command(root, args, cwd=target)
+                if result.returncode != 0:
+                    detail = result.stderr.strip() or result.stdout.strip() or "git baseline commit failed"
+                    failures.append(Failure("daily-execution-cli", f"`{label}` setup failed: {detail}"))
+                    return False
+            return True
+
+        prepare_review_target(review_target, "review run positive chain")
+        write_fake_codex(fake_bin / "codex", mode="success")
+        success_env = prepend_path_env(fake_bin)
+
+        payload, error = load_command_json(
+            root,
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "review",
+                "run",
+                "--target",
+                str(review_target),
+                "--item",
+                "INIT-0001",
+            ],
+            env=success_env,
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`review run` positive chain failed: {error}"))
+        else:
+            require_review_run_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`review run` positive chain",
+                payload=payload,
+                expected_result={"pass"},
+            )
+
+        engine_missing_target = Path(tmp) / "engine-missing"
+        prepare_review_target(engine_missing_target, "review run engine unavailable")
+        payload, error = load_command_json(
+            root,
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "review",
+                "run",
+                "--target",
+                str(engine_missing_target),
+                "--item",
+                "INIT-0001",
+            ],
+            env={"PATH": "/usr/bin:/bin"},
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`review run` engine unavailable failed: {error}"))
+        elif payload.get("result") != "block":
+            failures.append(Failure("daily-execution-cli", "`review run` must block when the default engine is unavailable"))
+        else:
+            require_review_run_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`review run` engine unavailable",
+                payload=payload,
+                expected_result={"block"},
+            )
+            engine = payload.get("engine")
+            if isinstance(engine, dict) and engine.get("failure_reason") != "engine_unavailable":
+                failures.append(Failure("daily-execution-cli", "`review run` must report `engine_unavailable` when Codex is missing"))
+            if payload.get("fallback_to") is not None:
+                failures.append(Failure("daily-execution-cli", "`review run` must not convert engine failure into checkpoint fallback"))
+
+        schema_target = Path(tmp) / "schema-drift"
+        prepare_review_target(schema_target, "review run schema drift")
+        write_fake_codex(fake_bin / "codex", mode="schema_drift")
+        payload, error = load_command_json(
+            root,
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "review",
+                "run",
+                "--target",
+                str(schema_target),
+                "--item",
+                "INIT-0001",
+            ],
+            env=success_env,
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`review run` schema drift failed: {error}"))
+        elif payload.get("result") != "block":
+            failures.append(Failure("daily-execution-cli", "`review run` must block on schema drift"))
+        else:
+            require_review_run_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`review run` schema drift",
+                payload=payload,
+                expected_result={"block"},
+            )
+            engine = payload.get("engine")
+            if isinstance(engine, dict) and engine.get("failure_reason") != "schema_drift":
+                failures.append(Failure("daily-execution-cli", "`review run` must report `schema_drift` for invalid engine output"))
+
+        dirty_target = Path(tmp) / "tracked-edit"
+        prepare_review_target(dirty_target, "review run tracked edit")
+        write_fake_codex(fake_bin / "codex", mode="tracked_edit", tracked_edit_target=".loom/status/current.md")
+        payload, error = load_command_json(
+            root,
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "review",
+                "run",
+                "--target",
+                str(dirty_target),
+                "--item",
+                "INIT-0001",
+            ],
+            env=success_env,
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`review run` tracked edit failed: {error}"))
+        elif payload.get("result") != "block":
+            failures.append(Failure("daily-execution-cli", "`review run` must block when engine modifies tracked repo content"))
+        else:
+            require_review_run_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`review run` tracked edit",
+                payload=payload,
+                expected_result={"block"},
+            )
+            engine = payload.get("engine")
+            if isinstance(engine, dict) and engine.get("failure_reason") != "repo_diff_detected":
+                failures.append(Failure("daily-execution-cli", "`review run` must report `repo_diff_detected` when tracked files change"))
+
+    with tempfile.TemporaryDirectory(prefix="loom-check-flow-") as tmp:
+        lifecycle_target = Path(tmp) / "new-project"
+        shutil.copytree(example_target, lifecycle_target)
+        temp_root = lifecycle_target / ".loom/flow/tmp"
+        temp_root.mkdir(parents=True, exist_ok=True)
+        (temp_root / "sentinel.txt").write_text("temp\n", encoding="utf-8")
+
+        for operation in ("create", "cleanup", "retire"):
+            payload, error = load_command_json(
+                root,
+                [
+                    "python3",
+                    "tools/loom_flow.py",
+                    "workspace",
+                    operation,
+                    "--target",
+                    str(lifecycle_target),
+                    "--item",
+                    "INIT-0001",
+                ],
+            )
+            if error:
+                failures.append(Failure("daily-execution-cli", f"`workspace {operation}` failed: {error}"))
+                continue
+            if payload.get("result") != "pass":
+                failures.append(
+                    Failure(
+                        "daily-execution-cli",
+                        f"`workspace {operation}` must pass on a clean temp copy, got `{payload.get('result')}`",
+                    )
+                )
+
+        locate_payload, error = load_command_json(
+            root,
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "workspace",
+                "locate",
+                "--target",
+                str(lifecycle_target),
+                "--item",
+                "INIT-0001",
+            ],
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`workspace locate` after retire failed: {error}"))
+        elif (
+            not isinstance(locate_payload.get("checkpoint"), dict)
+            or locate_payload["checkpoint"].get("normalized") != "retired"
+        ):
+            failures.append(Failure("daily-execution-cli", "`workspace retire` must leave the copied sample in `retired` state"))
+
+    with tempfile.TemporaryDirectory(prefix="loom-check-authoring-") as tmp:
+        authoring_target = Path(tmp) / "new-project"
+        shutil.copytree(example_target, authoring_target)
+
+        payload, error = load_command_json(
+            root,
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "recovery",
+                "writeback",
+                "--target",
+                str(authoring_target),
+                "--item",
+                "INIT-0001",
+                "--current-stop",
+                "Bootstrap review has started.",
+                "--next-step",
+                "Record the first formal review conclusion.",
+                "--latest-validation-summary",
+                "Bootstrap artifacts verified and ready for semantic review.",
+            ],
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`recovery writeback` failed: {error}"))
+        elif payload.get("result") != "pass":
+            failures.append(Failure("daily-execution-cli", "`recovery writeback` must pass on a clean temp copy"))
+
+        payload, error = load_command_json(
+            root,
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "work-item",
+                "create",
+                "--target",
+                str(authoring_target),
+                "--item",
+                "NEXT-0001",
+                "--goal",
+                "Validate work item authoring",
+                "--scope",
+                "Limit changes to `.loom/` artifacts for this temp check",
+                "--execution-path",
+                "execution/support",
+                "--workspace-entry",
+                ".",
+                "--validation-entry",
+                "python3 .loom/bin/loom_init.py verify --target .",
+                "--closing-condition",
+                "The authored work item can be activated and read mechanically.",
+                "--init-recovery",
+                "--activate",
+            ],
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`work-item create` failed: {error}"))
+        elif payload.get("result") != "pass":
+            failures.append(Failure("daily-execution-cli", "`work-item create --activate` must pass on a clean temp copy"))
+
+        payload, error = load_command_json(
+            root,
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "work-item",
+                "update",
+                "--target",
+                str(authoring_target),
+                "--item",
+                "NEXT-0001",
+                "--scope",
+                "Keep the temp authoring check constrained to `.loom/` files",
+                "--add-artifact",
+                ".loom/reviews/NEXT-0001.json",
+            ],
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`work-item update` failed: {error}"))
+        elif payload.get("result") != "pass":
+            failures.append(Failure("daily-execution-cli", "`work-item update` must pass on a clean temp copy"))
+
+        findings_path = authoring_target / ".loom" / "review-findings.json"
+        findings_path.parent.mkdir(parents=True, exist_ok=True)
+        findings_path.write_text(
+            json.dumps(
+                [
+                    {
+                        "id": "compat-block-1",
+                        "summary": "Formal review has not approved the item yet.",
+                        "severity": "block",
+                        "rebuttal": None,
+                        "disposition": {
+                            "status": "rejected",
+                            "summary": "The finding remains open until the missing approval signal is resolved."
+                        },
+                    },
+                    {
+                        "id": "compat-warn-1",
+                        "summary": "Re-run formal review after the missing approval signal is resolved.",
+                        "severity": "warn",
+                        "rebuttal": "A follow-up review will be recorded after the blocking issue is resolved.",
+                        "disposition": {
+                            "status": "deferred",
+                            "summary": "This follow-up stays open until the next formal review."
+                        },
+                    },
+                ],
+                ensure_ascii=False,
+                indent=2,
+            )
+            + "\n",
+            encoding="utf-8",
+        )
+
+        payload, error = load_command_json(
+            root,
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "review",
+                "record",
+                "--target",
+                str(authoring_target),
+                "--item",
+                "NEXT-0001",
+                "--decision",
+                "fallback",
+                "--kind",
+                "code_review",
+                "--summary",
+                "Formal review has not approved the item yet.",
+                "--reviewer",
+                "loom-check",
+                "--fallback-to",
+                "admission",
+                "--findings-file",
+                ".loom/review-findings.json",
+            ],
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`review record` failed: {error}"))
+        elif payload.get("result") != "pass":
+            failures.append(Failure("daily-execution-cli", "`review record` must pass for an authored fallback decision"))
+        else:
+            review = payload.get("review")
+            if isinstance(review, dict):
+                require_review_record_contract(
+                    failures,
+                    category="daily-execution-cli",
+                    context="`review record` review.record",
+                    payload=review.get("record"),
+                )
+
+    if shutil.which("git") is not None:
+        with tempfile.TemporaryDirectory(prefix="loom-check-purity-") as tmp:
+            dirty_target = Path(tmp) / "new-project"
+            shutil.copytree(example_target, dirty_target)
+            run_command(root, ["git", "init"], cwd=dirty_target)
+            run_command(root, ["git", "config", "user.email", "loom-check@example.com"], cwd=dirty_target)
+            run_command(root, ["git", "config", "user.name", "loom-check"], cwd=dirty_target)
+            run_command(root, ["git", "add", "."], cwd=dirty_target)
+            run_command(root, ["git", "commit", "-m", "baseline"], cwd=dirty_target)
+            (dirty_target / "untriaged.txt").write_text("pending\n", encoding="utf-8")
+            payload, error = load_command_json(
+                root,
+                [
+                    "python3",
+                    "tools/loom_flow.py",
+                    "purity-check",
+                    "--target",
+                    str(dirty_target),
+                    "--item",
+                    "INIT-0001",
+                ],
+            )
+            if error:
+                failures.append(Failure("daily-execution-cli", f"`purity-check` negative sample failed: {error}"))
+            elif payload.get("result") != "block":
+                failures.append(
+                    Failure(
+                        "daily-execution-cli",
+                        f"`purity-check` negative sample must block, got `{payload.get('result')}`",
+                    )
+                )
+            state_payload, error = load_command_json(
+                root,
+                [
+                    "python3",
+                    "tools/loom_flow.py",
+                    "state-check",
+                    "--target",
+                    str(dirty_target),
+                    "--item",
+                    "INIT-0001",
+                ],
+            )
+            if error:
+                failures.append(Failure("daily-execution-cli", f"`state-check` negative sample failed: {error}"))
+            elif state_payload.get("result") != "block":
+                failures.append(
+                    Failure(
+                        "daily-execution-cli",
+                        f"`state-check` negative sample must block, got `{state_payload.get('result')}`",
+                    )
+                )
+
+    with tempfile.TemporaryDirectory(prefix="loom-check-runtime-state-") as tmp:
+        tmp_root = Path(tmp)
+        install_root = tmp_root / "installed" / "skills"
+        target_root = tmp_root / "target"
+        bootstrap_target = tmp_root / "bootstrapped-target"
+        shutil.copytree(root / "skills", install_root)
+        target_root.mkdir(parents=True, exist_ok=True)
+        shutil.copytree(example_target, bootstrap_target)
+
+        payload, error = load_command_json(
+            root,
+            ["python3", str(install_root / "loom-init" / "scripts" / "loom-init.py"), "runtime-state", "--target", str(target_root)],
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`installed loom-init runtime-state` failed: {error}"))
+        else:
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`installed loom-init runtime-state`",
+                payload=payload.get("runtime_state"),
+                expected_scene="installed-runtime",
+                expected_carrier="installed-skills-root",
+                allowed_results={"pass"},
+            )
+
+        payload, error = load_command_json(
+            root,
+            ["python3", str(install_root / "shared" / "scripts" / "loom_flow.py"), "runtime-state", "--target", str(target_root)],
+            env={"LOOM_RUNTIME_SCENE": "upgrade-rehearsal"},
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`installed loom-flow runtime-state -- rehearsal` failed: {error}"))
+        else:
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`installed loom-flow runtime-state -- rehearsal`",
+                payload=payload.get("runtime_state"),
+                expected_scene="upgrade-rehearsal",
+                expected_carrier="installed-skills-root",
+                allowed_results={"pass"},
+            )
+
+        broken_install = tmp_root / "broken-install" / "skills"
+        shutil.copytree(root / "skills", broken_install)
+        (broken_install / "shared" / "scripts" / "loom_flow.py").unlink()
+        payload, error = load_command_json(
+            root,
+            ["python3", str(broken_install / "loom-init" / "scripts" / "loom-init.py"), "runtime-state", "--target", str(target_root)],
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`installed runtime-state` missing shared runtime failed unexpectedly: {error}"))
+        elif payload.get("result") != "block":
+            failures.append(Failure("daily-execution-cli", "`installed runtime-state` must block when shared runtime is missing"))
+
+        drift_install = tmp_root / "drift-install" / "skills"
+        shutil.copytree(root / "skills", drift_install)
+        (drift_install / "install-layout.json").unlink()
+        payload, error = load_command_json(
+            root,
+            ["python3", str(drift_install / "loom-init" / "scripts" / "loom-init.py"), "runtime-state", "--target", str(target_root)],
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`installed runtime-state` missing install-layout failed unexpectedly: {error}"))
+        elif payload.get("result") != "block":
+            failures.append(Failure("daily-execution-cli", "`installed runtime-state` must block when install-layout is missing"))
+
+        payload, error = load_command_json(
+            root,
+            ["python3", str(install_root / "shared" / "scripts" / "loom_flow.py"), "runtime-state", "--target", str(target_root)],
+            env={"LOOM_RUNTIME_SCENE": "repo-local-demo"},
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`installed runtime-state` scene conflict failed unexpectedly: {error}"))
+        elif payload.get("result") != "block":
+            failures.append(Failure("daily-execution-cli", "`installed runtime-state` must block on scene/carrier conflict"))
+
+        payload, error = load_command_json(
+            root,
+            ["python3", ".loom/bin/loom_init.py", "runtime-state", "--target", "."],
+            cwd=bootstrap_target,
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`bootstrapped loom-init runtime-state` failed: {error}"))
+        else:
+            require_runtime_state_payload(
+                failures,
+                category="daily-execution-cli",
+                context="`bootstrapped loom-init runtime-state`",
+                payload=payload.get("runtime_state"),
+                expected_scene="installed-runtime",
+                expected_carrier="bootstrapped-target-runtime",
+                allowed_results={"pass"},
+            )
+
+        broken_bootstrap = tmp_root / "broken-bootstrapped-target"
+        shutil.copytree(example_target, broken_bootstrap)
+        manifest_path = broken_bootstrap / ".loom" / "bootstrap" / "manifest.json"
+        manifest = load_json_file(manifest_path)
+        if isinstance(manifest, dict):
+            artifacts = manifest.get("artifacts")
+            if isinstance(artifacts, list):
+                for artifact in artifacts:
+                    if isinstance(artifact, dict) and artifact.get("path") == ".loom/bin/runtime_state.py":
+                        artifact["source"] = "broken/source.py"
+            manifest_path.write_text(json.dumps(manifest, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
+        payload, error = load_command_json(
+            root,
+            ["python3", ".loom/bin/loom_init.py", "runtime-state", "--target", "."],
+            cwd=broken_bootstrap,
+        )
+        if error:
+            failures.append(Failure("daily-execution-cli", f"`bootstrapped runtime-state` manifest drift failed unexpectedly: {error}"))
+        elif payload.get("result") != "block":
+            failures.append(Failure("daily-execution-cli", "`bootstrapped runtime-state` must block when the bootstrap manifest drifts"))
+
+    if shutil.which("git") is not None:
+        with tempfile.TemporaryDirectory(prefix="loom-check-installed-pre-merge-") as tmp:
+            tmp_root = Path(tmp)
+            install_root = tmp_root / "installed" / "skills"
+            source_snapshot = tmp_root / "source-snapshot"
+            positive_target = tmp_root / "positive-target"
+            review_fallback_target = tmp_root / "review-fallback-target"
+            fake_bin = tmp_root / "bin"
+            fake_bin.mkdir(parents=True, exist_ok=True)
+            write_fake_codex(fake_bin / "codex", mode="success")
+            installed_review_env = prepend_path_env(fake_bin)
+            shutil.copytree(root / "skills", install_root)
+            shutil.copytree(root, source_snapshot, ignore=shutil.ignore_patterns(".git", ".DS_Store", "__pycache__"))
+
+            def prepare_target(target: Path) -> tuple[str | None, list[str]]:
+                errors: list[str] = []
+                shutil.copytree(source_snapshot, target)
+                for args in (
+                    ["git", "init"],
+                    ["git", "config", "user.email", "loom-check@example.com"],
+                    ["git", "config", "user.name", "loom-check"],
+                ):
+                    result = run_command(root, args, cwd=target)
+                    if result.returncode != 0:
+                        detail = result.stderr.strip() or result.stdout.strip() or "git setup failed"
+                        errors.append(detail)
+                        return None, errors
+
+                payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "loom-init" / "scripts" / "loom-init.py"),
+                        "bootstrap",
+                        "--target",
+                        str(target),
+                        "--write",
+                        "--force",
+                        "--verify",
+                        "--install-pr-template",
+                    ],
+                )
+                if error:
+                    errors.append(error)
+                    return None, errors
+                verification = payload.get("verification")
+                if not isinstance(verification, dict) or verification.get("ok") is not True:
+                    errors.append("installed bootstrap must verify successfully before the pre-merge chain starts")
+                    return None, errors
+
+                git_add = run_command(root, ["git", "add", "."], cwd=target)
+                if git_add.returncode != 0:
+                    detail = git_add.stderr.strip() or git_add.stdout.strip() or "git add failed"
+                    errors.append(detail)
+                    return None, errors
+                git_commit = run_command(root, ["git", "commit", "-m", "bootstrap baseline for #209"], cwd=target)
+                if git_commit.returncode != 0:
+                    detail = git_commit.stderr.strip() or git_commit.stdout.strip() or "git commit failed"
+                    errors.append(detail)
+                    return None, errors
+
+                resume_payload, resume_error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "loom-resume" / "scripts" / "loom-resume.py"),
+                        "flow",
+                        "resume",
+                        "--target",
+                        str(target),
+                        "--item",
+                        "INIT-0001",
+                    ],
+                )
+                if resume_error:
+                    errors.append(resume_error)
+                    return None, errors
+                recovery = resume_payload.get("recovery")
+                if not isinstance(recovery, dict):
+                    errors.append("resume payload must include `recovery`")
+                    return None, errors
+                summary = recovery.get("latest_validation_summary")
+                if not isinstance(summary, str) or not summary:
+                    errors.append("resume payload must expose a non-empty `latest_validation_summary`")
+                    return None, errors
+                return summary, errors
+
+            positive_summary, positive_setup_errors = prepare_target(positive_target)
+            if positive_setup_errors:
+                failures.append(
+                    Failure(
+                        "daily-execution-cli",
+                        f"`installed pre-merge chain` setup failed: {'; '.join(positive_setup_errors)}",
+                    )
+                )
+            else:
+                task_signals = {
+                    "resume": "请接手当前事项并恢复上下文后继续推进",
+                    "pre-review": "请在进入 review 前做统一检查",
+                    "review": "请对当前事项做正式 review 并给出审查结论",
+                    "merge-ready": "请做 merge-ready 最终放行前预检并确认是否可以合并",
+                }
+
+                payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "loom-init" / "scripts" / "loom-init.py"),
+                        "route",
+                        "--target",
+                        str(positive_target),
+                        "--task",
+                        task_signals["resume"],
+                    ],
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed route resume` failed: {error}"))
+                else:
+                    require_route_payload(
+                        failures,
+                        category="daily-execution-cli",
+                        context="`installed route resume`",
+                        payload=payload,
+                        expected_skill="loom-resume",
+                        expected_mode="implicit",
+                        expected_runtime_scene="installed-runtime",
+                        expected_runtime_carrier="installed-skills-root",
+                    )
+
+                resume_payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "loom-resume" / "scripts" / "loom-resume.py"),
+                        "flow",
+                        "resume",
+                        "--target",
+                        str(positive_target),
+                        "--item",
+                        "INIT-0001",
+                    ],
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed flow resume` failed: {error}"))
+                elif resume_payload.get("result") != "pass":
+                    failures.append(Failure("daily-execution-cli", "`installed flow resume` must pass for the positive chain"))
+                else:
+                    require_runtime_state_payload(
+                        failures,
+                        category="daily-execution-cli",
+                        context="`installed flow resume`",
+                        payload=resume_payload.get("runtime_state"),
+                        expected_scene="installed-runtime",
+                        expected_carrier="installed-skills-root",
+                        allowed_results={"pass"},
+                    )
+
+                payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "loom-init" / "scripts" / "loom-init.py"),
+                        "route",
+                        "--target",
+                        str(positive_target),
+                        "--task",
+                        task_signals["pre-review"],
+                    ],
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed route pre-review` failed: {error}"))
+                else:
+                    require_route_payload(
+                        failures,
+                        category="daily-execution-cli",
+                        context="`installed route pre-review`",
+                        payload=payload,
+                        expected_skill="loom-pre-review",
+                        expected_mode="implicit",
+                        expected_runtime_scene="installed-runtime",
+                        expected_runtime_carrier="installed-skills-root",
+                    )
+
+                payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "loom-pre-review" / "scripts" / "loom-pre-review.py"),
+                        "flow",
+                        "pre-review",
+                        "--target",
+                        str(positive_target),
+                        "--item",
+                        "INIT-0001",
+                    ],
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed flow pre-review` failed: {error}"))
+                elif payload.get("result") != "pass":
+                    failures.append(Failure("daily-execution-cli", "`installed flow pre-review` must pass for the positive chain"))
+
+                payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "loom-init" / "scripts" / "loom-init.py"),
+                        "route",
+                        "--target",
+                        str(positive_target),
+                        "--task",
+                        task_signals["review"],
+                    ],
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed route review` failed: {error}"))
+                else:
+                    require_route_payload(
+                        failures,
+                        category="daily-execution-cli",
+                        context="`installed route review`",
+                        payload=payload,
+                        expected_skill="loom-review",
+                        expected_mode="implicit",
+                        expected_runtime_scene="installed-runtime",
+                        expected_runtime_carrier="installed-skills-root",
+                    )
+
+                review_flow_payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "loom-review" / "scripts" / "loom-review.py"),
+                        "flow",
+                        "review",
+                        "--target",
+                        str(positive_target),
+                        "--item",
+                        "INIT-0001",
+                    ],
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed flow review` failed: {error}"))
+                elif review_flow_payload.get("result") != "pass":
+                    failures.append(Failure("daily-execution-cli", "`installed flow review` must pass for the positive chain"))
+                else:
+                    review = review_flow_payload.get("review")
+                    if isinstance(review, dict):
+                        require_review_record_contract(
+                            failures,
+                            category="daily-execution-cli",
+                            context="`installed flow review` review.record",
+                            payload=review.get("record"),
+                        )
+
+                review_run_payload: dict[str, object] | None = None
+                review_record_input: dict[str, object] | None = None
+                review_run_payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "shared" / "scripts" / "loom_flow.py"),
+                        "review",
+                        "run",
+                        "--target",
+                        str(positive_target),
+                        "--item",
+                        "INIT-0001",
+                    ],
+                    env=installed_review_env,
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed review run` failed: {error}"))
+                elif review_run_payload.get("result") != "pass":
+                    failures.append(Failure("daily-execution-cli", "`installed review run` must pass for the positive chain"))
+                else:
+                    require_review_run_payload(
+                        failures,
+                        category="daily-execution-cli",
+                        context="`installed review run`",
+                        payload=review_run_payload,
+                        expected_result={"pass"},
+                    )
+                    review_record_input = review_run_payload.get("review_record_input") if isinstance(review_run_payload, dict) else None
+                review_record_payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "loom-review" / "scripts" / "loom-review.py"),
+                        "review",
+                        "record",
+                        "--target",
+                        str(positive_target),
+                        "--item",
+                        "INIT-0001",
+                        "--decision",
+                        str(review_record_input.get("decision", "allow")) if isinstance(review_record_input, dict) else "allow",
+                        "--kind",
+                        str(review_record_input.get("kind", "code_review")) if isinstance(review_record_input, dict) else "code_review",
+                        "--summary",
+                        str(review_record_input.get("summary", "Installed pre-merge chain is ready for merge checkpoint consumption."))
+                        if isinstance(review_record_input, dict)
+                        else "Installed pre-merge chain is ready for merge checkpoint consumption.",
+                        "--reviewer",
+                        str(review_record_input.get("reviewer", "loom-check")) if isinstance(review_record_input, dict) else "loom-check",
+                        "--findings-file",
+                        str(review_record_input.get("findings_file", ".loom/review-findings.json")) if isinstance(review_record_input, dict) else ".loom/review-findings.json",
+                        "--engine-adapter",
+                        str(review_record_input.get("engine_adapter", "loom/default-codex")) if isinstance(review_record_input, dict) else "loom/default-codex",
+                        "--engine-evidence",
+                        str(review_record_input.get("engine_evidence", ".loom/runtime/review/INIT-0001/unknown-head/engine-result.json"))
+                        if isinstance(review_record_input, dict)
+                        else ".loom/runtime/review/INIT-0001/unknown-head/engine-result.json",
+                        "--normalized-findings",
+                        str(review_record_input.get("normalized_findings", ".loom/review-findings.json"))
+                        if isinstance(review_record_input, dict)
+                        else ".loom/review-findings.json",
+                    ],
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed review record allow` failed: {error}"))
+                elif review_record_payload.get("result") != "pass":
+                    failures.append(Failure("daily-execution-cli", "`installed review record allow` must pass"))
+                else:
+                    review = review_record_payload.get("review")
+                    if isinstance(review, dict):
+                        require_review_record_contract(
+                            failures,
+                            category="daily-execution-cli",
+                            context="`installed review record allow` review.record",
+                            payload=review.get("record"),
+                        )
+
+                payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "shared" / "scripts" / "loom_flow.py"),
+                        "recovery",
+                        "writeback",
+                        "--target",
+                        str(positive_target),
+                        "--item",
+                        "INIT-0001",
+                        "--current-checkpoint",
+                        "merge checkpoint",
+                        "--current-stop",
+                        "Installed review completed and merge-ready validation is next.",
+                        "--next-step",
+                        "Run merge-ready and checkpoint merge from installed skills.",
+                        "--latest-validation-summary",
+                        positive_summary,
+                    ],
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed recovery writeback for merge` failed: {error}"))
+                elif payload.get("result") != "pass":
+                    failures.append(Failure("daily-execution-cli", "`installed recovery writeback for merge` must pass"))
+
+                git_add = run_command(
+                    root,
+                    [
+                        "git",
+                        "add",
+                        "-f",
+                        ".loom/progress/INIT-0001.md",
+                        ".loom/status/current.md",
+                        ".loom/reviews/INIT-0001.json",
+                    ],
+                    cwd=positive_target,
+                )
+                if git_add.returncode != 0:
+                    detail = git_add.stderr.strip() or git_add.stdout.strip() or "git add failed"
+                    failures.append(Failure("daily-execution-cli", f"`installed pre-merge carrier commit` add failed: {detail}"))
+                else:
+                    git_commit = run_command(
+                        root,
+                        ["git", "commit", "-m", "author installed pre-merge carriers for #209"],
+                        cwd=positive_target,
+                    )
+                    if git_commit.returncode != 0:
+                        detail = git_commit.stderr.strip() or git_commit.stdout.strip() or "git commit failed"
+                        failures.append(Failure("daily-execution-cli", f"`installed pre-merge carrier commit` failed: {detail}"))
+
+                payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "loom-init" / "scripts" / "loom-init.py"),
+                        "route",
+                        "--target",
+                        str(positive_target),
+                        "--task",
+                        task_signals["merge-ready"],
+                    ],
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed route merge-ready` failed: {error}"))
+                else:
+                    require_route_payload(
+                        failures,
+                        category="daily-execution-cli",
+                        context="`installed route merge-ready`",
+                        payload=payload,
+                        expected_skill="loom-merge-ready",
+                        expected_mode="implicit",
+                        expected_runtime_scene="installed-runtime",
+                        expected_runtime_carrier="installed-skills-root",
+                    )
+
+                merge_ready_payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "loom-merge-ready" / "scripts" / "loom-merge-ready.py"),
+                        "flow",
+                        "merge-ready",
+                        "--target",
+                        str(positive_target),
+                        "--item",
+                        "INIT-0001",
+                    ],
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed flow merge-ready` failed: {error}"))
+                elif merge_ready_payload.get("result") != "pass":
+                    failures.append(Failure("daily-execution-cli", "`installed flow merge-ready` must pass for the positive chain"))
+                else:
+                    merge_checkpoint = merge_ready_payload.get("merge_checkpoint")
+                    if not isinstance(merge_checkpoint, dict) or merge_checkpoint.get("result") != "pass":
+                        failures.append(Failure("daily-execution-cli", "`installed flow merge-ready` must expose `merge_checkpoint.result = pass`"))
+
+                checkpoint_merge_payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "shared" / "scripts" / "loom_flow.py"),
+                        "checkpoint",
+                        "merge",
+                        "--target",
+                        str(positive_target),
+                        "--item",
+                        "INIT-0001",
+                    ],
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed checkpoint merge` failed: {error}"))
+                elif checkpoint_merge_payload.get("result") != "pass":
+                    failures.append(Failure("daily-execution-cli", "`installed checkpoint merge` must pass for the positive chain"))
+
+                broken_install = tmp_root / "broken-install" / "skills"
+                shutil.copytree(root / "skills", broken_install)
+                (broken_install / "install-layout.json").unlink()
+                payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(broken_install / "loom-init" / "scripts" / "loom-init.py"),
+                        "route",
+                        "--target",
+                        str(positive_target),
+                        "--task",
+                        task_signals["resume"],
+                    ],
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed route` missing install-layout failed unexpectedly: {error}"))
+                else:
+                    require_route_payload(
+                        failures,
+                        category="daily-execution-cli",
+                        context="`installed route` missing install-layout",
+                        payload=payload,
+                        expected_skill="loom-init",
+                        expected_mode="fallback",
+                        expected_runtime_scene="installed-runtime",
+                        expected_runtime_carrier="installed-skills-root",
+                        allowed_results={"block"},
+                    )
+
+                payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(broken_install / "loom-pre-review" / "scripts" / "loom-pre-review.py"),
+                        "flow",
+                        "pre-review",
+                        "--target",
+                        str(positive_target),
+                        "--item",
+                        "INIT-0001",
+                    ],
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed flow pre-review` missing install-layout failed unexpectedly: {error}"))
+                elif payload.get("result") != "block":
+                    failures.append(Failure("daily-execution-cli", "`installed flow pre-review` must block when install-layout is missing"))
+                else:
+                    require_runtime_state_payload(
+                        failures,
+                        category="daily-execution-cli",
+                        context="`installed flow pre-review` missing install-layout",
+                        payload=payload.get("runtime_state"),
+                        expected_scene="installed-runtime",
+                        expected_carrier="installed-skills-root",
+                        allowed_results={"block"},
+                    )
+
+                review_fallback_summary, review_fallback_errors = prepare_target(review_fallback_target)
+                if review_fallback_errors:
+                    failures.append(
+                        Failure(
+                            "daily-execution-cli",
+                            f"`installed review baseline fallback` setup failed: {'; '.join(review_fallback_errors)}",
+                        )
+                    )
+                else:
+                    payload, error = load_command_json(
+                        root,
+                        [
+                            "python3",
+                            str(install_root / "shared" / "scripts" / "loom_flow.py"),
+                            "recovery",
+                            "writeback",
+                            "--target",
+                            str(review_fallback_target),
+                            "--item",
+                            "INIT-0001",
+                            "--current-checkpoint",
+                            "admission checkpoint",
+                            "--current-stop",
+                            "Installed review baseline is still at admission.",
+                            "--next-step",
+                            "Promote the target repo to build checkpoint before review.",
+                            "--latest-validation-summary",
+                            review_fallback_summary,
+                        ],
+                    )
+                    if error:
+                        failures.append(Failure("daily-execution-cli", f"`installed recovery writeback for admission fallback` failed: {error}"))
+                    elif payload.get("result") != "pass":
+                        failures.append(Failure("daily-execution-cli", "`installed recovery writeback for admission fallback` must pass"))
+
+                    git_add = run_command(
+                        root,
+                        ["git", "add", "-f", ".loom/progress/INIT-0001.md", ".loom/status/current.md"],
+                        cwd=review_fallback_target,
+                    )
+                    if git_add.returncode != 0:
+                        detail = git_add.stderr.strip() or git_add.stdout.strip() or "git add failed"
+                        failures.append(Failure("daily-execution-cli", f"`installed review baseline fallback` add failed: {detail}"))
+                    else:
+                        git_commit = run_command(
+                            root,
+                            ["git", "commit", "-m", "lower checkpoint to admission for #209 fallback"],
+                            cwd=review_fallback_target,
+                        )
+                        if git_commit.returncode != 0:
+                            detail = git_commit.stderr.strip() or git_commit.stdout.strip() or "git commit failed"
+                            failures.append(Failure("daily-execution-cli", f"`installed review baseline fallback` commit failed: {detail}"))
+
+                    payload, error = load_command_json(
+                        root,
+                        [
+                            "python3",
+                            str(install_root / "loom-review" / "scripts" / "loom-review.py"),
+                            "flow",
+                            "review",
+                            "--target",
+                            str(review_fallback_target),
+                            "--item",
+                            "INIT-0001",
+                        ],
+                    )
+                    if error:
+                        failures.append(Failure("daily-execution-cli", f"`installed flow review` admission fallback failed: {error}"))
+                    elif payload.get("result") != "fallback" or payload.get("fallback_to") != "admission":
+                        failures.append(Failure("daily-execution-cli", "`installed flow review` must fall back to `admission` when build checkpoint is missing"))
+
+                    payload, error = load_command_json(
+                        root,
+                        [
+                            "python3",
+                            str(install_root / "loom-merge-ready" / "scripts" / "loom-merge-ready.py"),
+                            "flow",
+                            "merge-ready",
+                            "--target",
+                            str(review_fallback_target),
+                            "--item",
+                            "INIT-0001",
+                        ],
+                    )
+                    if error:
+                        failures.append(Failure("daily-execution-cli", f"`installed flow merge-ready` review-baseline fallback failed: {error}"))
+                    elif payload.get("result") not in {"fallback", "block"}:
+                        failures.append(Failure("daily-execution-cli", "`installed flow merge-ready` must fail closed when review baseline is missing"))
+
+                readme_path = positive_target / "README.md"
+                readme_path.write_text(readme_path.read_text(encoding="utf-8") + "\n# review-head-drift\n", encoding="utf-8")
+                git_add = run_command(root, ["git", "add", "README.md"], cwd=positive_target)
+                if git_add.returncode != 0:
+                    detail = git_add.stderr.strip() or git_add.stdout.strip() or "git add failed"
+                    failures.append(Failure("daily-execution-cli", f"`installed merge-ready drift` add failed: {detail}"))
+                else:
+                    git_commit = run_command(
+                        root,
+                        ["git", "commit", "-m", "introduce non-carrier drift after review for #209"],
+                        cwd=positive_target,
+                    )
+                    if git_commit.returncode != 0:
+                        detail = git_commit.stderr.strip() or git_commit.stdout.strip() or "git commit failed"
+                        failures.append(Failure("daily-execution-cli", f"`installed merge-ready drift` commit failed: {detail}"))
+
+                payload, error = load_command_json(
+                    root,
+                    [
+                        "python3",
+                        str(install_root / "shared" / "scripts" / "loom_flow.py"),
+                        "checkpoint",
+                        "merge",
+                        "--target",
+                        str(positive_target),
+                        "--item",
+                        "INIT-0001",
+                    ],
+                )
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`installed checkpoint merge` drift negative failed: {error}"))
+                elif payload.get("result") != "block":
+                    failures.append(Failure("daily-execution-cli", "`installed checkpoint merge` must block when HEAD drifts beyond Loom carriers"))
+
+    gh_auth_ready = shutil.which("gh") is not None and run_command(root, ["gh", "auth", "status"]).returncode == 0
+    if gh_auth_ready:
+        with tempfile.TemporaryDirectory(prefix="loom-check-installed-post-merge-") as tmp:
+            tmp_root = Path(tmp)
+            install_root = tmp_root / "installed" / "skills"
+            retire_target = tmp_root / "retire-target"
+            dirty_target = tmp_root / "dirty-target"
+            broken_install = tmp_root / "broken-install" / "skills"
+            shutil.copytree(root / "skills", install_root)
+
+            for label, args in (
+                (
+                    "installed reconciliation audit",
+                    [
+                        "python3",
+                        str(install_root / "shared" / "scripts" / "loom_flow.py"),
+                        "reconciliation",
+                        "audit",
+                        "--target",
+                        str(root),
+                        "--issue",
+                        "131",
+                        "--pr",
+                        "138",
+                        "--project",
+                        "5",
+                    ],
+                ),
+                (
+                    "installed reconciliation sync dry-run",
+                    [
+                        "python3",
+                        str(install_root / "shared" / "scripts" / "loom_flow.py"),
+                        "reconciliation",
+                        "sync",
+                        "--target",
+                        str(root),
+                        "--issue",
+                        "131",
+                        "--pr",
+                        "138",
+                        "--project",
+                        "5",
+                        "--dry-run",
+                    ],
+                ),
+                (
+                    "installed closeout check",
+                    [
+                        "python3",
+                        str(install_root / "shared" / "scripts" / "loom_flow.py"),
+                        "closeout",
+                        "check",
+                        "--target",
+                        str(root),
+                        "--issue",
+                        "131",
+                        "--pr",
+                        "138",
+                        "--project",
+                        "5",
+                        "--skip-gate",
+                    ],
+                ),
+                (
+                    "installed closeout sync",
+                    [
+                        "python3",
+                        str(install_root / "shared" / "scripts" / "loom_flow.py"),
+                        "closeout",
+                        "sync",
+                        "--target",
+                        str(root),
+                        "--issue",
+                        "131",
+                        "--pr",
+                        "138",
+                        "--project",
+                        "5",
+                        "--skip-gate",
+                    ],
+                ),
+            ):
+                payload, error = load_command_json_with_retry(
+                    root,
+                    args,
+                    timeout_seconds=30,
+                    retries=3,
+                )
+                if error:
+                    if label in {"installed closeout check", "installed closeout sync"} and "command timed out" in error:
+                        continue
+                    failures.append(Failure("daily-execution-cli", f"`{label}` failed: {error}"))
+                    continue
+                rate_limited = payload_has_github_rate_limit(payload)
+                if label == "installed reconciliation audit":
+                    if payload.get("result") != "pass" and not rate_limited:
+                        failures.append(Failure("daily-execution-cli", "`installed reconciliation audit` must pass on the historical closeout sample"))
+                    require_runtime_state_payload(
+                        failures,
+                        category="daily-execution-cli",
+                        context="`installed reconciliation audit`",
+                        payload=payload.get("runtime_state"),
+                        expected_scene="installed-runtime",
+                        expected_carrier="installed-skills-root",
+                        allowed_results={"pass"},
+                    )
+                    require_reconciliation_payload(
+                        failures,
+                        category="daily-execution-cli",
+                        context="`installed reconciliation audit`",
+                        payload=payload,
+                    )
+                elif label == "installed reconciliation sync dry-run":
+                    if payload.get("result") != "pass" and not rate_limited:
+                        failures.append(Failure("daily-execution-cli", "`installed reconciliation sync --dry-run` must pass on an already aligned sample"))
+                    require_runtime_state_payload(
+                        failures,
+                        category="daily-execution-cli",
+                        context="`installed reconciliation sync --dry-run`",
+                        payload=payload.get("runtime_state"),
+                        expected_scene="installed-runtime",
+                        expected_carrier="installed-skills-root",
+                        allowed_results={"pass"},
+                    )
+                else:
+                    if payload.get("result") != "pass" and not rate_limited:
+                        failures.append(Failure("daily-execution-cli", f"`{label}` must pass on the historical closeout sample"))
+                    require_runtime_state_payload(
+                        failures,
+                        category="daily-execution-cli",
+                        context=f"`{label}`",
+                        payload=payload.get("runtime_state"),
+                        expected_scene="installed-runtime",
+                        expected_carrier="installed-skills-root",
+                        allowed_results={"pass"},
+                    )
+                    require_closeout_reconciliation_contract(
+                        failures,
+                        category="daily-execution-cli",
+                        context=f"`{label}`",
+                        payload=payload,
+                    )
+
+            for target in (retire_target, dirty_target):
+                shutil.copytree(example_target, target)
+                for args in (
+                    ["git", "init"],
+                    ["git", "config", "user.email", "loom-check@example.com"],
+                    ["git", "config", "user.name", "loom-check"],
+                    ["git", "add", "."],
+                    ["git", "commit", "-m", "baseline"],
+                ):
+                    result = run_command(root, args, cwd=target)
+                    if result.returncode != 0:
+                        detail = result.stderr.strip() or result.stdout.strip() or "git setup failed"
+                        failures.append(Failure("daily-execution-cli", f"`installed retire` setup failed: {detail}"))
+                        break
+
+            purity_payload, error = load_command_json(
+                root,
+                [
+                    "python3",
+                    str(install_root / "loom-retire" / "scripts" / "loom-retire.py"),
+                    "purity-check",
+                    "--target",
+                    str(retire_target),
+                    "--item",
+                    "INIT-0001",
+                ],
+            )
+            if error:
+                failures.append(Failure("daily-execution-cli", f"`installed purity-check` failed: {error}"))
+            elif purity_payload.get("result") != "pass":
+                failures.append(Failure("daily-execution-cli", "`installed purity-check` must pass on a clean retire target"))
+            else:
+                require_runtime_state_payload(
+                    failures,
+                    category="daily-execution-cli",
+                    context="`installed purity-check`",
+                    payload=purity_payload.get("runtime_state"),
+                    expected_scene="installed-runtime",
+                    expected_carrier="installed-skills-root",
+                    allowed_results={"pass"},
+                )
+
+            temp_root = retire_target / ".loom" / ".tmp"
+            temp_root.mkdir(parents=True, exist_ok=True)
+            (temp_root / "sentinel.txt").write_text("temp\n", encoding="utf-8")
+
+            cleanup_payload, error = load_command_json(
+                root,
+                [
+                    "python3",
+                    str(install_root / "loom-retire" / "scripts" / "loom-retire.py"),
+                    "workspace",
+                    "cleanup",
+                    "--target",
+                    str(retire_target),
+                    "--item",
+                    "INIT-0001",
+                ],
+            )
+            if error:
+                failures.append(Failure("daily-execution-cli", f"`installed workspace cleanup` failed: {error}"))
+            elif cleanup_payload.get("result") != "pass":
+                failures.append(Failure("daily-execution-cli", "`installed workspace cleanup` must pass for Loom-owned residue"))
+            else:
+                require_runtime_state_payload(
+                    failures,
+                    category="daily-execution-cli",
+                    context="`installed workspace cleanup`",
+                    payload=cleanup_payload.get("runtime_state"),
+                    expected_scene="installed-runtime",
+                    expected_carrier="installed-skills-root",
+                    allowed_results={"pass"},
+                )
+
+            retire_payload, error = load_command_json(
+                root,
+                [
+                    "python3",
+                    str(install_root / "loom-retire" / "scripts" / "loom-retire.py"),
+                    "workspace",
+                    "retire",
+                    "--target",
+                    str(retire_target),
+                    "--item",
+                    "INIT-0001",
+                ],
+            )
+            if error:
+                failures.append(Failure("daily-execution-cli", f"`installed workspace retire` failed: {error}"))
+            elif retire_payload.get("result") != "pass":
+                failures.append(Failure("daily-execution-cli", "`installed workspace retire` must pass after cleanup"))
+            else:
+                require_runtime_state_payload(
+                    failures,
+                    category="daily-execution-cli",
+                    context="`installed workspace retire`",
+                    payload=retire_payload.get("runtime_state"),
+                    expected_scene="installed-runtime",
+                    expected_carrier="installed-skills-root",
+                    allowed_results={"pass"},
+                )
+                checkpoint = retire_payload.get("checkpoint")
+                if not isinstance(checkpoint, dict) or checkpoint.get("normalized") != "retired":
+                    failures.append(Failure("daily-execution-cli", "`installed workspace retire` must leave the target in `retired` state"))
+
+            (dirty_target / "foreign-residue.txt").write_text("pending\n", encoding="utf-8")
+            dirty_add = run_command(root, ["git", "add", "foreign-residue.txt"], cwd=dirty_target)
+            if dirty_add.returncode != 0:
+                detail = dirty_add.stderr.strip() or dirty_add.stdout.strip() or "git add failed"
+                failures.append(Failure("daily-execution-cli", f"`installed retire` dirty sample setup failed: {detail}"))
+
+            for label, args in (
+                (
+                    "installed purity-check dirty sample",
+                    [
+                        "python3",
+                        str(install_root / "loom-retire" / "scripts" / "loom-retire.py"),
+                        "purity-check",
+                        "--target",
+                        str(dirty_target),
+                        "--item",
+                        "INIT-0001",
+                    ],
+                ),
+                (
+                    "installed workspace cleanup dirty sample",
+                    [
+                        "python3",
+                        str(install_root / "loom-retire" / "scripts" / "loom-retire.py"),
+                        "workspace",
+                        "cleanup",
+                        "--target",
+                        str(dirty_target),
+                        "--item",
+                        "INIT-0001",
+                    ],
+                ),
+                (
+                    "installed workspace retire dirty sample",
+                    [
+                        "python3",
+                        str(install_root / "loom-retire" / "scripts" / "loom-retire.py"),
+                        "workspace",
+                        "retire",
+                        "--target",
+                        str(dirty_target),
+                        "--item",
+                        "INIT-0001",
+                    ],
+                ),
+            ):
+                payload, error = load_command_json(root, args)
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`{label}` failed: {error}"))
+                    continue
+                if payload.get("result") != "block":
+                    failures.append(Failure("daily-execution-cli", f"`{label}` must block when non-Loom residue is present"))
+                require_runtime_state_payload(
+                    failures,
+                    category="daily-execution-cli",
+                    context=f"`{label}`",
+                    payload=payload.get("runtime_state"),
+                    expected_scene="installed-runtime",
+                    expected_carrier="installed-skills-root",
+                    allowed_results={"pass"},
+                )
+
+            shutil.copytree(root / "skills", broken_install)
+            (broken_install / "install-layout.json").unlink()
+            for label, args in (
+                (
+                    "installed closeout check missing install-layout",
+                    [
+                        "python3",
+                        str(broken_install / "shared" / "scripts" / "loom_flow.py"),
+                        "closeout",
+                        "check",
+                        "--target",
+                        str(root),
+                        "--issue",
+                        "131",
+                        "--pr",
+                        "138",
+                        "--skip-gate",
+                    ],
+                ),
+                (
+                    "installed purity-check missing install-layout",
+                    [
+                        "python3",
+                        str(broken_install / "loom-retire" / "scripts" / "loom-retire.py"),
+                        "purity-check",
+                        "--target",
+                        str(retire_target),
+                        "--item",
+                        "INIT-0001",
+                    ],
+                ),
+            ):
+                payload, error = load_command_json(root, args)
+                if error:
+                    failures.append(Failure("daily-execution-cli", f"`{label}` failed unexpectedly: {error}"))
+                    continue
+                if payload.get("result") != "block":
+                    failures.append(Failure("daily-execution-cli", f"`{label}` must block when install-layout is missing"))
+                require_runtime_state_payload(
+                    failures,
+                    category="daily-execution-cli",
+                    context=f"`{label}`",
+                    payload=payload.get("runtime_state"),
+                    expected_scene="installed-runtime",
+                    expected_carrier="installed-skills-root",
+                    allowed_results={"block"},
+                )
+
+    fail_closed_payloads = [
+        (
+            "closeout-fix-needed-fail-open",
+            {
+                "result": "pass",
+                "fallback_to": None,
+                "reconciliation": {
+                    "command": "reconciliation",
+                    "operation": "audit",
+                    "result": "fix-needed",
+                    "summary": "fix-needed",
+                    "missing_inputs": [],
+                    "fallback_to": "manual-reconciliation",
+                    "findings": [
+                        {
+                            "kind": "absorbed_but_open",
+                            "severity": "fix-needed",
+                            "subject": "issue #177",
+                            "evidence": {},
+                            "recommended_action": "run reconciliation sync",
+                        }
+                    ],
+                },
+            },
+        ),
+        (
+            "closeout-block-fallback-drift",
+            {
+                "result": "block",
+                "fallback_to": "merge",
+                "reconciliation": {
+                    "command": "reconciliation",
+                    "operation": "audit",
+                    "result": "block",
+                    "summary": "block",
+                    "missing_inputs": ["issue/pr/project"],
+                    "fallback_to": "manual-reconciliation",
+                    "findings": [
+                        {
+                            "kind": "parent_drift",
+                            "severity": "block",
+                            "subject": "parent issue #148",
+                            "evidence": {},
+                            "recommended_action": "manual reconciliation",
+                        }
+                    ],
+                },
+            },
+        ),
+        (
+            "closeout-malformed-reconciliation",
+            {
+                "result": "pass",
+                "fallback_to": None,
+                "reconciliation": {
+                    "command": "reconciliation",
+                    "operation": "audit",
+                    "summary": "broken",
+                    "missing_inputs": "bad",
+                    "findings": "bad",
+                },
+            },
+        ),
+    ]
+    for label, payload in fail_closed_payloads:
+        sample_failures: list[Failure] = []
+        require_closeout_reconciliation_contract(
+            sample_failures,
+            category="daily-execution-cli",
+            context=f"`{label}`",
+            payload=payload,
+        )
+        if not sample_failures:
+            failures.append(
+                Failure(
+                    "daily-execution-cli",
+                    f"`{label}` synthetic payload must fail closeout reconciliation validation",
+                )
+            )
+
+    warn_payload_failures: list[Failure] = []
+    require_closeout_reconciliation_contract(
+        warn_payload_failures,
+        category="daily-execution-cli",
+        context="`closeout-warn-does-not-block`",
+        payload={
+            "result": "pass",
+            "fallback_to": None,
+            "reconciliation": {
+                "command": "reconciliation",
+                "operation": "audit",
+                "result": "warn",
+                "summary": "warn",
+                "missing_inputs": [],
+                "fallback_to": "manual-reconciliation",
+                "findings": [
+                    {
+                        "kind": "project_drift",
+                        "severity": "warn",
+                        "subject": "project 5",
+                        "evidence": {},
+                        "recommended_action": "review warning",
+                    }
+                ],
+            },
+        },
+    )
+    if warn_payload_failures:
+        failures.append(
+            Failure(
+                "daily-execution-cli",
+                "`closeout-warn-does-not-block` synthetic payload must allow non-blocking reconciliation warnings",
+            )
+        )
+
+    return failures
+
+
+def check_repo_companion_interface_contracts(root: Path) -> list[Failure]:
+    failures: list[Failure] = []
+    example_target = root / "examples/new-project"
+    if not example_target.exists():
+        return failures
+
+    def write_json(path: Path, payload: object) -> None:
+        path.parent.mkdir(parents=True, exist_ok=True)
+        path.write_text(json.dumps(payload, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
+
+    def install_companion(
+        target: Path,
+        *,
+        manifest: dict[str, object] | None = None,
+        repo_interface: dict[str, object] | None = None,
+        legacy_docs_only: bool = False,
+    ) -> None:
+        companion_dir = target / ".loom" / "companion"
+        companion_dir.mkdir(parents=True, exist_ok=True)
+        if legacy_docs_only:
+            (companion_dir / "README.md").write_text("# Legacy Companion Docs\n", encoding="utf-8")
+            return
+        (companion_dir / "README.md").write_text("# Repo Companion\n", encoding="utf-8")
+        for doc in (
+            "review.md",
+            "merge-ready.md",
+            "closeout.md",
+            "specialized-gates.md",
+            "checkpoints.md",
+            "metadata-contract.md",
+            "context-schema.md",
+        ):
+            (companion_dir / doc).write_text(f"# {doc}\n", encoding="utf-8")
+        if manifest is not None:
+            write_json(companion_dir / "manifest.json", manifest)
+        if repo_interface is not None:
+            write_json(companion_dir / "repo-interface.json", repo_interface)
+
+    valid_manifest = {
+        "schema_version": "loom-repo-companion-manifest/v1",
+        "companion_entry": ".loom/companion/README.md",
+        "repo_interface": ".loom/companion/repo-interface.json",
+    }
+    valid_interface_v1 = {
+        "schema_version": "loom-repo-interface/v1",
+        "companion_entry": ".loom/companion/README.md",
+        "repo_specific_requirements": {
+            "review": [
+                {
+                    "id": "review-specialized-gate",
+                    "summary": "Run the repo-specific semantic review checklist.",
+                    "locator": ".loom/companion/review.md",
+                    "enforcement": "blocking",
+                }
+            ],
+            "merge_ready": [
+                {
+                    "id": "merge-ready-advisory-note",
+                    "summary": "Review the repo-specific merge advisory note.",
+                    "locator": ".loom/companion/merge-ready.md",
+                    "enforcement": "advisory",
+                }
+            ],
+            "closeout": [
+                {
+                    "id": "closeout-specialized-gate",
+                    "summary": "Confirm the repo-specific closeout checklist.",
+                    "locator": ".loom/companion/closeout.md",
+                    "enforcement": "blocking",
+                }
+            ],
+        },
+        "specialized_gates": [
+            {
+                "id": "specialized-release-gate",
+                "summary": "Companion-owned release judgment.",
+                "locator": ".loom/companion/specialized-gates.md",
+            }
+        ],
+    }
+    valid_interface_v2 = {
+        "schema_version": "loom-repo-interface/v2",
+        "companion_entry": ".loom/companion/README.md",
+        "repo_specific_requirements": valid_interface_v1["repo_specific_requirements"],
+        "specialized_gates": [
+            {
+                "id": "specialized-review-gate",
+                "summary": "Companion-owned review specialization.",
+                "locator": ".loom/companion/specialized-gates.md",
+                "gate_type": "review",
+            }
+        ],
+        "metadata_contract": {
+            "fields": [
+                {
+                    "id": "integration_check",
+                    "summary": "Declare repo-specific integration metadata.",
+                    "applicability_locator": ".loom/companion/metadata-contract.md",
+                    "authority_locator": ".loom/companion/review.md",
+                    "enforcement": "blocking",
+                }
+            ]
+        },
+        "context_schema": {
+            "fields": [
+                {
+                    "id": "item_key",
+                    "summary": "Repo-native item key.",
+                    "type": "string",
+                    "required": True,
+                    "mapping_rule_locator": ".loom/companion/context-schema.md",
+                }
+            ]
+        },
+    }
+
+    with tempfile.TemporaryDirectory(prefix="loom-check-repo-companion-") as tmp:
+        base = Path(tmp)
+
+        absent_target = base / "absent"
+        shutil.copytree(example_target, absent_target)
+        absent_surface = build_governance_surface(absent_target)
+        repo_interface = absent_surface.get("repo_interface")
+        require_repo_interface_payload(
+            failures,
+            category="repo-companion",
+            context="absent repo companion",
+            payload=repo_interface,
+        )
+        if not isinstance(repo_interface, dict) or repo_interface.get("availability") != "absent":
+            failures.append(Failure("repo-companion", "absent repo companion sample must report `availability: absent`"))
+
+        docs_only_target = base / "docs-only"
+        shutil.copytree(example_target, docs_only_target)
+        install_companion(docs_only_target, legacy_docs_only=True)
+        docs_only_surface = build_governance_surface(docs_only_target)
+        docs_only_interface = docs_only_surface.get("repo_interface")
+        require_repo_interface_payload(
+            failures,
+            category="repo-companion",
+            context="docs-only repo companion",
+            payload=docs_only_interface,
+        )
+        if not isinstance(docs_only_interface, dict) or docs_only_interface.get("availability") != "companion_docs_only":
+            failures.append(Failure("repo-companion", "docs-only repo companion sample must report `availability: companion_docs_only`"))
+
+        incomplete_target = base / "incomplete"
+        shutil.copytree(example_target, incomplete_target)
+        install_companion(
+            incomplete_target,
+            manifest={
+                **valid_manifest,
+                "current_stop": "forbidden authored state",
+                "repo_interface": ".loom/companion/missing-interface.json",
+            },
+        )
+        incomplete_surface = build_governance_surface(incomplete_target)
+        incomplete_interface = incomplete_surface.get("repo_interface")
+        require_repo_interface_payload(
+            failures,
+            category="repo-companion",
+            context="incomplete repo companion",
+            payload=incomplete_interface,
+        )
+        if not isinstance(incomplete_interface, dict) or incomplete_interface.get("availability") != "incomplete":
+            failures.append(Failure("repo-companion", "incomplete repo companion sample must report `availability: incomplete`"))
+
+        invalid_interface_target = base / "invalid-interface"
+        shutil.copytree(example_target, invalid_interface_target)
+        install_companion(
+            invalid_interface_target,
+            manifest=valid_manifest,
+            repo_interface={
+                "schema_version": "loom-repo-interface/v1",
+                "companion_entry": ".loom/companion/README.md",
+                "repo_specific_requirements": {
+                    "review": [
+                        {
+                            "id": "bad-enforcement",
+                            "summary": "Broken requirement",
+                            "locator": ".loom/companion/review.md",
+                            "enforcement": "required",
+                        }
+                    ],
+                    "merge_ready": [],
+                },
+                "specialized_gates": [],
+            },
+        )
+        invalid_interface_surface = build_governance_surface(invalid_interface_target)
+        invalid_interface = invalid_interface_surface.get("repo_interface")
+        require_repo_interface_payload(
+            failures,
+            category="repo-companion",
+            context="invalid repo companion interface",
+            payload=invalid_interface,
+        )
+        if not isinstance(invalid_interface, dict) or invalid_interface.get("availability") != "incomplete":
+            failures.append(Failure("repo-companion", "invalid repo companion interface sample must report `availability: incomplete`"))
+
+        invalid_v2_target = base / "invalid-v2-interface"
+        shutil.copytree(example_target, invalid_v2_target)
+        install_companion(
+            invalid_v2_target,
+            manifest=valid_manifest,
+            repo_interface={
+                "schema_version": "loom-repo-interface/v2",
+                "companion_entry": ".loom/companion/README.md",
+                "repo_specific_requirements": valid_interface_v1["repo_specific_requirements"],
+                "specialized_gates": [
+                    {
+                        "id": "bad-gate-type",
+                        "summary": "Broken gate type",
+                        "locator": ".loom/companion/specialized-gates.md",
+                        "gate_type": "guardian",
+                    }
+                ],
+                "metadata_contract": {
+                    "fields": [
+                        {
+                            "id": "bad-metadata",
+                            "summary": "Broken metadata field",
+                            "applicability_locator": ".loom/companion/metadata-contract.md",
+                            "authority_locator": ".loom/companion/review.md",
+                            "enforcement": "required",
+                        }
+                    ]
+                },
+                "context_schema": {
+                    "fields": [
+                        {
+                            "id": "bad-context",
+                            "summary": "Broken context field",
+                            "type": "object",
+                            "required": "yes",
+                            "mapping_rule_locator": ".loom/companion/context-schema.md",
+                        }
+                    ]
+                },
+            },
+        )
+        invalid_v2_surface = build_governance_surface(invalid_v2_target)
+        invalid_v2_interface = invalid_v2_surface.get("repo_interface")
+        require_repo_interface_payload(
+            failures,
+            category="repo-companion",
+            context="invalid v2 repo companion interface",
+            payload=invalid_v2_interface,
+        )
+        if not isinstance(invalid_v2_interface, dict) or invalid_v2_interface.get("availability") != "incomplete":
+            failures.append(Failure("repo-companion", "invalid v2 repo companion interface sample must report `availability: incomplete`"))
+
+        present_v1_target = base / "present-v1"
+        shutil.copytree(example_target, present_v1_target)
+        install_companion(
+            present_v1_target,
+            manifest=valid_manifest,
+            repo_interface=valid_interface_v1,
+        )
+        present_v1_surface = build_governance_surface(present_v1_target)
+        present_v1_interface = present_v1_surface.get("repo_interface")
+        require_repo_interface_payload(
+            failures,
+            category="repo-companion",
+            context="present v1 repo companion",
+            payload=present_v1_interface,
+        )
+        if not isinstance(present_v1_interface, dict) or present_v1_interface.get("availability") != "present":
+            failures.append(Failure("repo-companion", "present v1 repo companion sample must report `availability: present`"))
+
+        present_target = base / "present-v2"
+        shutil.copytree(example_target, present_target)
+        install_companion(
+            present_target,
+            manifest=valid_manifest,
+            repo_interface=valid_interface_v2,
+        )
+        present_surface = build_governance_surface(present_target)
+        present_interface = present_surface.get("repo_interface")
+        require_repo_interface_payload(
+            failures,
+            category="repo-companion",
+            context="present v2 repo companion",
+            payload=present_interface,
+        )
+        if not isinstance(present_interface, dict) or present_interface.get("availability") != "present":
+            failures.append(Failure("repo-companion", "present v2 repo companion sample must report `availability: present`"))
+
+        review_requirements = repo_specific_requirements_payload(
+            present_interface,
+            target_root=present_target,
+            surface="review",
+        )
+        require_repo_specific_requirements_payload(
+            failures,
+            category="repo-companion",
+            context="present repo companion review requirements",
+            payload=review_requirements,
+            expected_surface="review",
+        )
+        if review_requirements.get("result") != "block":
+            failures.append(Failure("repo-companion", "blocking review requirements must fail closed"))
+
+        merge_requirements = repo_specific_requirements_payload(
+            present_interface,
+            target_root=present_target,
+            surface="merge_ready",
+        )
+        require_repo_specific_requirements_payload(
+            failures,
+            category="repo-companion",
+            context="present repo companion merge-ready requirements",
+            payload=merge_requirements,
+            expected_surface="merge_ready",
+        )
+        if merge_requirements.get("result") != "pass":
+            failures.append(Failure("repo-companion", "advisory merge-ready requirements must remain non-blocking"))
+
+        closeout_requirements = repo_specific_requirements_payload(
+            present_interface,
+            target_root=present_target,
+            surface="closeout",
+        )
+        require_repo_specific_requirements_payload(
+            failures,
+            category="repo-companion",
+            context="present repo companion closeout requirements",
+            payload=closeout_requirements,
+            expected_surface="closeout",
+        )
+        if closeout_requirements.get("result") != "block":
+            failures.append(Failure("repo-companion", "blocking closeout requirements must fail closed"))
+
+        flow_review_payload, error = load_command_json(
+            root,
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "flow",
+                "review",
+                "--target",
+                str(present_target),
+                "--item",
+                "INIT-0001",
+            ],
+        )
+        if error:
+            failures.append(Failure("repo-companion", f"`flow review` companion sample failed: {error}"))
+        elif flow_review_payload.get("result") != "block":
+            failures.append(Failure("repo-companion", "`flow review` must block when repo companion declares blocking review requirements"))
+
+        flow_merge_ready_payload, error = load_command_json(
+            root,
+            [
+                "python3",
+                "tools/loom_flow.py",
+                "flow",
+                "merge-ready",
+                "--target",
+                str(present_target),
+                "--item",
+                "INIT-0001",
+            ],
+        )
+        if error:
+            failures.append(Failure("repo-companion", f"`flow merge-ready` companion sample failed: {error}"))
+        elif not isinstance(flow_merge_ready_payload.get("repo_specific_requirements"), dict):
+            failures.append(Failure("repo-companion", "`flow merge-ready` companion sample must include `repo_specific_requirements`"))
+        elif flow_merge_ready_payload["repo_specific_requirements"].get("result") != "pass":
+            failures.append(Failure("repo-companion", "`flow merge-ready` advisory companion requirements must stay non-blocking"))
+
+    return failures
+
+
+def check_repo_interop_contracts(root: Path) -> list[Failure]:
+    example_target = root / "examples/new-project"
+    if not example_target.exists():
+        return []
+
+    failures: list[Failure] = []
+
+    def write_json(path: Path, payload: object) -> None:
+        path.parent.mkdir(parents=True, exist_ok=True)
+        path.write_text(json.dumps(payload, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
+
+    def install_interop(
+        target: Path,
+        *,
+        interop: dict[str, object] | None = None,
+    ) -> None:
+        companion_dir = target / ".loom" / "companion"
+        companion_dir.mkdir(parents=True, exist_ok=True)
+        (target / "host").mkdir(parents=True, exist_ok=True)
+        (target / "native").mkdir(parents=True, exist_ok=True)
+        (target / ".loom" / "shadow").mkdir(parents=True, exist_ok=True)
+        (target / "native" / "status").mkdir(parents=True, exist_ok=True)
+        for relative, payload in {
+            ".loom/shadow/admission-loom.json": {"result": "pass"},
+            ".loom/shadow/admission-repo.json": {"result": "pass"},
+            ".loom/shadow/review-loom.json": {"decision": "allow"},
+            ".loom/shadow/review-repo.json": {"decision": "allow"},
+            ".loom/shadow/merge-ready-loom.json": {"status": "pass"},
+            ".loom/shadow/merge-ready-repo.json": {"status": "pass"},
+            ".loom/shadow/closeout-loom.json": {"status": "done"},
+            ".loom/shadow/closeout-repo.json": {"status": "done"},
+            "host/guardian-review.json": {"verdict": "allow"},
+        }.items():
+            write_json(target / relative, payload)
+        if interop is not None:
+            write_json(companion_dir / "interop.json", interop)
+
+    valid_interop = {
+        "schema_version": "loom-repo-interop/v1",
+        "host_adapters": [
+            {
+                "id": "guardian-review",
+                "summary": "Read guardian review verdicts without reimplementing the host action.",
+                "surfaces": ["review", "merge_ready"],
+                "locator": "host/guardian-review.json",
+            }
+        ],
+        "repo_native_carriers": [
+            {
+                "id": "governance-status",
+                "summary": "Read repo-native governance status output without migrating carriers.",
+                "surfaces": ["admission", "review", "merge_ready", "closeout"],
+                "locator": "native/status",
+            }
+        ],
+        "shadow_surfaces": {
+            "admission": {
+                "summary": "Compare admission parity.",
+                "loom_locator": ".loom/shadow/admission-loom.json",
+                "repo_locator": ".loom/shadow/admission-repo.json",
+            },
+            "review": {
+                "summary": "Compare review parity.",
+                "loom_locator": ".loom/shadow/review-loom.json",
+                "repo_locator": ".loom/shadow/review-repo.json",
+            },
+            "merge_ready": {
+                "summary": "Compare merge-ready parity.",
+                "loom_locator": ".loom/shadow/merge-ready-loom.json",
+                "repo_locator": ".loom/shadow/merge-ready-repo.json",
+            },
+            "closeout": {
+                "summary": "Compare closeout parity.",
+                "loom_locator": ".loom/shadow/closeout-loom.json",
+                "repo_locator": ".loom/shadow/closeout-repo.json",
+            },
+        },
+    }
+
+    with tempfile.TemporaryDirectory(prefix="loom-check-repo-interop-") as tmp:
+        base = Path(tmp)
+
+        absent_target = base / "absent"
+        shutil.copytree(example_target, absent_target)
+        absent_surface = build_governance_surface(absent_target)
+        repo_interop = absent_surface.get("repo_interop")
+        require_repo_interop_payload(
+            failures,
+            category="repo-interop",
+            context="absent repo interop",
+            payload=repo_interop,
+        )
+        if not isinstance(repo_interop, dict) or repo_interop.get("availability") != "absent":
+            failures.append(Failure("repo-interop", "absent repo interop sample must report `availability: absent`"))
+
+        invalid_target = base / "invalid"
+        shutil.copytree(example_target, invalid_target)
+        install_interop(
+            invalid_target,
+            interop={
+                "schema_version": "loom-repo-interop/v1",
+                "host_adapters": [
+                    {
+                        "id": "bad-adapter",
+                        "summary": "Broken adapter",
+                        "surfaces": ["guardian"],
+                        "locator": "host/missing.json",
+                    }
+                ],
+                "repo_native_carriers": [],
+                "shadow_surfaces": {
+                    "admission": {
+                        "summary": "Compare admission parity.",
+                        "loom_locator": ".loom/shadow/admission-loom.json",
+                        "repo_locator": ".loom/shadow/admission-repo.json",
+                    }
+                },
+            },
+        )
+        invalid_surface = build_governance_surface(invalid_target)
+        invalid_interop = invalid_surface.get("repo_interop")
+        require_repo_interop_payload(
+            failures,
+            category="repo-interop",
+            context="invalid repo interop",
+            payload=invalid_interop,
+        )
+        if not isinstance(invalid_interop, dict) or invalid_interop.get("availability") != "incomplete":
+            failures.append(Failure("repo-interop", "invalid repo interop sample must report `availability: incomplete`"))
+
+        present_target = base / "present"
+        shutil.copytree(example_target, present_target)
+        install_interop(present_target, interop=valid_interop)
+        present_surface = build_governance_surface(present_target)
+        present_interop = present_surface.get("repo_interop")
+        require_repo_interop_payload(
+            failures,
+            category="repo-interop",
+            context="present repo interop",
+            payload=present_interop,
+        )
+        if not isinstance(present_interop, dict) or present_interop.get("availability") != "present":
+            failures.append(Failure("repo-interop", "present repo interop sample must report `availability: present`"))
+
+        parity_payload, error = load_command_json(
+            root,
+            ["python3", "tools/loom_flow.py", "shadow-parity", "--target", str(present_target)],
+        )
+        if error:
+            failures.append(Failure("repo-interop", f"`shadow-parity` sample failed: {error}"))
+        else:
+            require_shadow_parity_payload(
+                failures,
+                category="repo-interop",
+                context="`shadow-parity` present sample",
+                payload=parity_payload,
+                expected_reports=4,
+            )
+            if parity_payload.get("result") != "pass":
+                failures.append(Failure("repo-interop", "`shadow-parity` must pass when all declared surfaces match"))
+
+        mismatch_target = base / "mismatch"
+        shutil.copytree(present_target, mismatch_target)
+        write_json(mismatch_target / ".loom/shadow/review-repo.json", {"decision": "block"})
+        mismatch_payload, error = load_command_json(
+            root,
+            ["python3", "tools/loom_flow.py", "shadow-parity", "--target", str(mismatch_target), "--surface", "review"],
+        )
+        if error:
+            failures.append(Failure("repo-interop", f"`shadow-parity` mismatch sample failed: {error}"))
+        else:
+            require_shadow_parity_payload(
+                failures,
+                category="repo-interop",
+                context="`shadow-parity` mismatch sample",
+                payload=mismatch_payload,
+                expected_reports=1,
+            )
+            reports = mismatch_payload.get("reports")
+            if not isinstance(reports, list) or not reports or reports[0].get("result") != "mismatch":
+                failures.append(Failure("repo-interop", "`shadow-parity` mismatch sample must report `mismatch`"))
+
+    return failures
+
+
+def is_within(path: Path, root: Path) -> bool:
+    try:
+        path.relative_to(root)
+        return True
+    except ValueError:
+        return False
+
+
+def collect_failures(root: Path) -> list[Failure]:
+    failures: list[Failure] = []
+    failures.extend(check_required_paths(root, "top-level-dirs", TOP_LEVEL_DIRS))
+    failures.extend(check_required_paths(root, "top-level-files", TOP_LEVEL_FILES))
+    failures.extend(check_required_paths(root, "area-readmes", AREA_READMES))
+    failures.extend(check_required_paths(root, "core-docs", CORE_DOCS))
+    failures.extend(
+        check_required_paths(root, "automation-frontload-templates", AUTOMATION_FRONTLOAD_TEMPLATES)
+    )
+    failures.extend(check_required_paths(root, "automation-frontload-skills", AUTOMATION_FRONTLOAD_SKILLS))
+    failures.extend(
+        check_required_paths(
+            root,
+            "automation-frontload-execution-support",
+            AUTOMATION_FRONTLOAD_EXECUTION_SUPPORT,
+        )
+    )
+    failures.extend(check_skill_manifests(root))
+    failures.extend(check_skill_routing(root))
+    failures.extend(check_demo_assets(root))
+    failures.extend(check_demo_fact_chain(root))
+    failures.extend(check_demo_repo_local_cli(root))
+    failures.extend(check_deep_existing_repo_bootstrap(root))
+    failures.extend(check_daily_execution_cli(root))
+    failures.extend(check_repo_companion_interface_contracts(root))
+    failures.extend(check_repo_interop_contracts(root))
+    failures.extend(check_markdown_links(root))
+    return failures
+
+
+def print_report(root: Path, failures: list[Failure]) -> None:
+    categories_checked = 15
+    if not failures:
+        print(f"loom_check: OK ({root})")
+        print(f"checked {categories_checked} surfaces")
+        return
+
+    grouped: dict[str, list[str]] = defaultdict(list)
+    for failure in failures:
+        grouped[failure.category].append(failure.detail)
+
+    print(f"loom_check: FAILED ({root})")
+    for category in sorted(grouped):
+        print(f"- {category}")
+        for detail in grouped[category]:
+            print(f"  - {detail}")
+    print(f"failures: {len(failures)} across {len(grouped)} categories")
+
+
+def main(argv: list[str]) -> int:
+    root = repo_root_from_argv(argv)
+    if not root.exists():
+        print(f"loom_check: repo root does not exist: {root}", file=sys.stderr)
+        return 2
+    failures = collect_failures(root)
+    print_report(root, failures)
+    return 1 if failures else 0
+
+
+if __name__ == "__main__":
+    sys.exit(main(sys.argv))