@clawpump/claw-agent 0.1.4 → 0.1.6

package/agent/agent/coding_context.py

@@ -0,0 +1,738 @@
+"""Coding-context awareness — base Hermes, every interactive surface.
+
+When the user runs Hermes inside a code workspace (CLI, TUI, desktop app, or an
+editor over ACP), Hermes shifts into a **coding posture**. This module is the
+single place that decides whether we're in that posture and what it implies,
+so the rest of the codebase never re-derives "are we coding?" on its own.
+
+Architecture — one seam, many consumers
+----------------------------------------
+The posture is modelled as a frozen :class:`RuntimeMode` selected from a small
+:class:`ContextProfile` registry (today: ``coding`` and ``general``). A profile
+is *data* — it declares the toolset to collapse to, the operating brief to
+inject, and hints for other domains (model routing, memory, subagents). Every
+domain reads the same resolved object instead of probing git/config itself:
+
+  * **System prompt** — ``RuntimeMode.system_blocks()`` → the operating brief +
+    a live git/workspace snapshot (``agent/system_prompt.py``).
+  * **Toolset** — ``RuntimeMode.toolset_selection()`` → the ``coding`` toolset
+    plus the user's enabled MCP servers (``cli.py`` / ``tui_gateway``). Only
+    under the opt-in ``focus`` mode: the default posture is prompt-only and
+    never touches the user's configured toolsets (toolsets like messaging /
+    smart-home / music are off-by-default anyway, and someone who explicitly
+    enabled image-gen or Spotify shouldn't lose it for being in a git repo).
+  * **Delegation** — subagents inherit the parent's toolset and run through the
+    same prompt builder, so the coding posture propagates to children for free.
+  * **Model / memory / compression** — declared on the profile
+    (``model_hint``, ``memory_policy``) as the extension seam; consumers read
+    ``mode.profile`` rather than re-deciding.
+
+Cache safety
+------------
+The mode is resolved **once** and is immutable. The workspace snapshot is built
+once at prompt-build time and baked into the *stable* system-prompt tier — never
+re-probed per turn (that would shatter the prompt cache). Branch and dirty state
+drift mid-session, so the brief tells the model to re-check with ``git`` before
+acting on the snapshot. A ``/coding`` flip therefore only takes effect next
+session (deferred), the same contract as ``/skills install`` vs ``--now``.
+
+Activation (config ``agent.coding_context``):
+
+  * ``auto`` (default) — posture (brief + snapshot) on an interactive coding
+    surface sitting in a code workspace (git repo or recognised project root).
+    Prompt-only; toolsets and the skill index untouched.
+  * ``focus`` — like ``auto``, but additionally collapses the toolset to the
+    ``coding`` set + enabled MCP servers and demotes non-coding skill
+    categories to names-only in the prompt's skill index (no skill is ever
+    hidden). Explicit opt-in for a lean schema.
+  * ``on`` — force the posture anywhere (incl. non-workspaces). Prompt-only.
+  * ``off`` — disable entirely.
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+import os
+import re
+import subprocess
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Any, Optional
+
+logger = logging.getLogger("hermes.coding_context")
+
+CODING_TOOLSET = "coding"
+
+# Surfaces where a coding posture makes sense under ``auto``. Messaging
+# platforms (telegram, discord, slack, …) are intentionally absent — a chat bot
+# in a group is not pair-programming.
+INTERACTIVE_CODING_PLATFORMS = {"cli", "tui", "acp", "desktop", ""}
+
+# Project-root signals that mark a directory as a code workspace even when it
+# isn't (yet) a git repo. Cheap filename checks — no parsing.
+_PROJECT_MARKERS = (
+    "pyproject.toml", "setup.py", "setup.cfg", "requirements.txt",
+    "package.json", "tsconfig.json", "deno.json",
+    "Cargo.toml", "go.mod", "pom.xml", "build.gradle", "build.gradle.kts",
+    "Gemfile", "composer.json", "mix.exs", "pubspec.yaml",
+    "CMakeLists.txt", "Makefile", "Dockerfile",
+    "AGENTS.md", "CLAUDE.md", ".cursorrules",
+)
+
+# Agent-instruction files surfaced separately from manifests in the snapshot.
+_CONTEXT_FILES = ("AGENTS.md", "CLAUDE.md", ".cursorrules")
+
+# Lockfile → package manager, checked in priority order.
+_PY_LOCKFILES = (("uv.lock", "uv"), ("poetry.lock", "poetry"), ("Pipfile.lock", "pipenv"))
+_JS_LOCKFILES = (
+    ("pnpm-lock.yaml", "pnpm"), ("bun.lockb", "bun"), ("bun.lock", "bun"),
+    ("yarn.lock", "yarn"), ("package-lock.json", "npm"),
+)
+
+# package.json scripts / Makefile targets worth surfacing as verify commands.
+_VERIFY_TARGETS = ("test", "tests", "lint", "typecheck", "check", "build", "fmt", "format")
+_MAX_VERIFY_COMMANDS = 8
+_MAX_FACT_FILE_BYTES = 256 * 1024
+
+_GIT_TIMEOUT = 2.5
+
+
+# Per-model edit-format steering. Matching the edit tool format to how a model
+# was trained reduces mistakes and wasted reasoning (OpenAI/Codex handle
+# patch-style diffs best; Anthropic models — and most open-weight coding
+# models, whose RL scaffolds use str_replace-style editors — do best with
+# string-replacement). Our `patch` tool exposes both: mode="patch" (V4A
+# multi-file) and mode="replace" (find-and-swap). We nudge each family toward
+# its native format. Unknown families get nothing (the brief's neutral wording
+# stands). Substrings match the model id; aligned with TOOL_USE_ENFORCEMENT_MODELS.
+#
+# GPT/Codex get V4A for ALL edits, single-file included: in codex-rs,
+# apply_patch (V4A — apply_patch.lark) is the ONLY file editor, no
+# str_replace-style tool exists, and the shipped model prompts say to use
+# apply_patch even "for single file edits" — so a replace-mode nudge would
+# steer those models toward a format their first-party harness never taught
+# them.
+_EDIT_FORMAT_GUIDANCE: dict[str, tuple[tuple[str, ...], str]] = {
+    "patch": (
+        ("gpt", "codex"),
+        "- Edit format: author new files with `write_file`; for edits to "
+        "existing code use `patch` with `mode='patch'` (V4A diff) — including "
+        "single-file edits. It's the edit format you handle most reliably.",
+    ),
+    "replace": (
+        ("claude", "sonnet", "opus", "haiku",
+         "gemini", "gemma", "deepseek", "qwen", "kimi", "glm", "grok",
+         "hermes", "llama", "mistral", "devstral", "minimax"),
+        "- Edit format: author new files with `write_file`; for edits to "
+        "existing code prefer `patch` in `mode='replace'` — match a unique "
+        "snippet and swap it. Reach for `mode='patch'` (V4A) only when an edit "
+        "genuinely spans several files at once.",
+    ),
+}
+
+
+def _model_family(model: Optional[str]) -> Optional[str]:
+    """Classify a model id into an edit-format family key, or ``None``.
+
+    Used to steer the coding posture toward the edit tool format a model was
+    trained on. Family-agnostic by design: an unrecognised model gets ``None``
+    and the operating brief's neutral edit wording applies.
+    """
+    if not model:
+        return None
+    lowered = model.lower()
+    for family, (needles, _line) in _EDIT_FORMAT_GUIDANCE.items():
+        if any(n in lowered for n in needles):
+            return family
+    return None
+
+
+def _edit_format_line(model: Optional[str]) -> str:
+    """The edit-format guidance line for this model's family (``""`` if none)."""
+    family = _model_family(model)
+    if family is None:
+        return ""
+    return _EDIT_FORMAT_GUIDANCE[family][1]
+
+
+# Operating brief for the coding posture. Tool names referenced here (read_file,
+# search_files, patch, write_file, terminal, todo) are in the coding toolset and
+# in _HERMES_CORE_TOOLS, so they're present on every surface this fires on.
+CODING_AGENT_GUIDANCE = (
+    "You are a coding agent pairing with the user inside their codebase. "
+    "Operate like a careful senior engineer.\n"
+    "\n"
+    "Gather context first:\n"
+    "- Read the relevant files with `read_file` and locate code with "
+    "`search_files` before changing anything. Trace a symbol to its definition "
+    "and usages rather than guessing its shape.\n"
+    "- Batch independent lookups: when several reads/searches don't depend on "
+    "each other, issue them together in one turn instead of one at a time.\n"
+    "- Never invent files, symbols, APIs, or imports. If you haven't seen it in "
+    "the repo, go look. Don't assume a library is available — check the project "
+    "manifest (pyproject.toml / package.json / Cargo.toml / go.mod) and how "
+    "neighbouring files import it.\n"
+    "\n"
+    "Make changes through the tools, not the chat:\n"
+    "- Edit with `patch`/`write_file`. Do NOT print code blocks to the user as "
+    "a substitute for editing — apply the change, then summarise it. Only show "
+    "code when the user explicitly asks to see it.\n"
+    "- Match the project's existing style and conventions; AGENTS.md / "
+    "CLAUDE.md / .cursorrules already in context win over your defaults. Touch "
+    "only what the task needs — no drive-by refactors, renames, or reformatting "
+    "— and add any imports/dependencies your code requires.\n"
+    "- If an edit fails to apply, re-read the file to get the current exact "
+    "contents before retrying — don't repeat a stale patch. If the same region "
+    "fails twice, rewrite the enclosing function or file with `write_file` "
+    "instead of attempting a third patch.\n"
+    "\n"
+    "Verify, and know when to stop:\n"
+    "- Use `terminal` for git, builds, tests, and inspection. Run the relevant "
+    "tests/linter/build and confirm they pass before claiming the work is done.\n"
+    "- Terminal state persists across calls: current directory and exported "
+    "environment variables carry forward. Activate a virtualenv or export setup "
+    "vars once, then reuse that state instead of re-sourcing it before every "
+    "test command.\n"
+    "- Fix root causes, not symptoms: when you find a bug, check sibling call "
+    "paths for the same flaw and fix the class, not just the reported site.\n"
+    "- When fixing linter/type errors on a file, stop after about three "
+    "attempts on the same file and ask the user rather than looping.\n"
+    "- Track multi-step work with `todo`. Reference code as `path:line` instead "
+    "of pasting whole files.\n"
+    "\n"
+    "Respect the user's repo: don't commit, push, or rewrite history unless "
+    "asked, and never read, print, or commit secrets — leave `.env` and "
+    "credential files alone unless the user explicitly asks. The Workspace "
+    "block below is a snapshot from session start — re-run `git status`/"
+    "`git branch` before relying on it. Be concise: lead with the change or "
+    "answer, not a preamble."
+)
+
+
+# ── Context profiles (declarative posture definitions) ──────────────────────
+
+
+@dataclass(frozen=True)
+class ContextProfile:
+    """A named operating posture. Pure data — consumers read these fields.
+
+    ``toolset``      — collapse to this toolset (+ enabled MCP) when no explicit
+                       selection is pinned; ``None`` keeps the platform default.
+    ``guidance``     — operating brief injected into the stable system prompt;
+                       ``""`` injects nothing.
+    ``model_hint``   — routing preference key for smart model routing
+                       (extension seam; not yet consumed by the router).
+    ``memory_policy``— memory namespace/weighting hint (extension seam).
+    ``compact_skill_categories`` — skill categories DEMOTED to names-only in
+                       the system-prompt skill index under the opt-in ``focus``
+                       mode. Never hidden: every skill name stays visible
+                       (so memory-anchored recall keeps working) — only the
+                       descriptions are dropped to cut index noise. Deny-list
+                       semantics so unknown/custom categories keep full
+                       entries.
+    """
+
+    name: str
+    toolset: Optional[str] = None
+    guidance: str = ""
+    model_hint: Optional[str] = None
+    memory_policy: str = "default"
+    compact_skill_categories: tuple[str, ...] = ()
+
+
+# Skill categories that are clearly not part of a coding workflow. Demoted to
+# names-only in the prompt's skill index under the opt-in ``focus`` mode only
+# (deny-list — anything not listed here, incl. custom user categories, keeps
+# full entries). Coding-adjacent categories (devops, github, mcp,
+# data-science, diagramming, research, security, …) are intentionally absent.
+_NON_CODING_SKILL_CATEGORIES = (
+    "apple", "communication", "cooking", "creative", "email", "finance",
+    "gaming", "gifs", "health", "media", "music", "note-taking",
+    "productivity", "shopping", "smart-home", "social-media", "travel",
+    "yuanbao",
+)
+
+
+GENERAL_PROFILE = ContextProfile(name="general")
+CODING_PROFILE = ContextProfile(
+    name="coding",
+    toolset=CODING_TOOLSET,
+    guidance=CODING_AGENT_GUIDANCE,
+    model_hint="coding",
+    memory_policy="project",
+    compact_skill_categories=_NON_CODING_SKILL_CATEGORIES,
+)
+
+_PROFILES: dict[str, ContextProfile] = {
+    GENERAL_PROFILE.name: GENERAL_PROFILE,
+    CODING_PROFILE.name: CODING_PROFILE,
+}
+
+
+def get_profile(name: str) -> ContextProfile:
+    """Return a registered profile, falling back to ``general``."""
+    return _PROFILES.get(name, GENERAL_PROFILE)
+
+
+# ── Helpers ─────────────────────────────────────────────────────────────────
+
+
+def _coding_mode(config: Optional[dict[str, Any]]) -> str:
+    """Return the normalized ``agent.coding_context`` mode (auto/focus/on/off)."""
+    if config is None:
+        try:
+            from hermes_cli.config import load_config
+
+            config = load_config()
+        except Exception:
+            config = {}
+    raw = ((config or {}).get("agent", {}) or {}).get("coding_context", "auto")
+    mode = str(raw).strip().lower()
+    if mode in {"focus", "strict", "lean"}:
+        return "focus"
+    if mode in {"on", "true", "yes", "1", "always"}:
+        return "on"
+    if mode in {"off", "false", "no", "0", "never"}:
+        return "off"
+    return "auto"
+
+
+def _resolve_cwd(cwd: Optional[str | Path]) -> Path:
+    if cwd:
+        return Path(cwd).expanduser()
+    try:
+        from agent.runtime_cwd import resolve_agent_cwd
+
+        return resolve_agent_cwd()
+    except Exception:
+        return Path(os.getcwd())
+
+
+def _git_root(cwd: Path) -> Optional[Path]:
+    current = cwd.resolve()
+    for parent in [current, *current.parents]:
+        if (parent / ".git").exists():
+            return parent
+    return None
+
+
+def _home() -> Optional[Path]:
+    try:
+        return Path.home().resolve()
+    except (OSError, RuntimeError):
+        return None
+
+
+def _marker_root(cwd: Path) -> Optional[Path]:
+    """Nearest ancestor that looks like a project root, or ``None``.
+
+    Walks up at most a few levels so a manifest in the workspace root counts
+    even when the user is in a subdirectory. ``$HOME`` itself is skipped — a
+    Makefile or AGENTS.md sitting in the home directory is global user config,
+    not a project-root signal.
+    """
+    current = cwd.resolve()
+    home = _home()
+    for depth, parent in enumerate([current, *current.parents]):
+        if depth > 6:
+            break
+        if parent == home:
+            continue
+        for marker in _PROJECT_MARKERS:
+            if (parent / marker).exists():
+                return parent
+    return None
+
+
+def _detect_profile_name(mode: str, platform: str, cwd_str: str) -> str:
+    """Resolve which profile applies.
+
+    ``auto``/``focus``: coding when the surface is interactive AND the cwd is a
+    code workspace (a git repo or a recognised project root). ``on``: always
+    coding. ``off``: always general.
+
+    A git repo rooted at ``$HOME`` (the dotfiles pattern) is NOT a workspace
+    signal — without the guard, every session anywhere under a dotfiles-managed
+    home directory would silently flip to the coding posture.
+
+    Detection is intentionally not memoized: it's a handful of ``stat`` calls,
+    and callers resolve the mode once per session anyway. Caching here would
+    risk a stale posture if a long-lived process (gateway/TUI) serves sessions
+    from different working directories.
+    """
+    if mode == "off":
+        return GENERAL_PROFILE.name
+    if mode == "on":
+        return CODING_PROFILE.name
+    if platform and platform.strip().lower() not in INTERACTIVE_CODING_PLATFORMS:
+        return GENERAL_PROFILE.name
+    cwd = Path(cwd_str)
+    git_root = _git_root(cwd)
+    if git_root is not None and git_root == _home():
+        git_root = None  # dotfiles repo at $HOME — not a code workspace
+    if git_root is not None or _marker_root(cwd) is not None:
+        return CODING_PROFILE.name
+    return GENERAL_PROFILE.name
+
+
+# ── RuntimeMode (the seam) ──────────────────────────────────────────────────
+
+
+@dataclass(frozen=True)
+class RuntimeMode:
+    """The resolved operating posture for a session. Immutable by construction.
+
+    Built once via :func:`resolve_runtime_mode` and consumed by every domain
+    that cares about the coding/general distinction. Never mutate or re-resolve
+    mid-session — that would break the prompt cache.
+    """
+
+    profile: ContextProfile
+    surface: str
+    cwd: Path
+    # The normalized ``agent.coding_context`` mode this posture was resolved
+    # under (auto/focus/on/off). Toolset collapse is gated on ``focus``.
+    config_mode: str = "auto"
+    # The model id this session runs (e.g. "anthropic/claude-opus-4.8"). Used
+    # only to steer edit-format guidance toward the model's family — see
+    # ``_edit_format_line``. Fixed for the session, so cache-safe.
+    model: Optional[str] = None
+
+    @property
+    def kind(self) -> str:
+        return self.profile.name
+
+    @property
+    def is_coding(self) -> bool:
+        return self.profile.name == CODING_PROFILE.name
+
+    def toolset_selection(self, config: Optional[dict[str, Any]] = None) -> Optional[list[str]]:
+        """Toolset list for this posture, or ``None`` to keep the platform default.
+
+        Non-``None`` only under the opt-in ``focus`` mode. The default posture
+        is prompt-only: most strippable toolsets are off-by-default anyway, and
+        a user who explicitly enabled one (image-gen for frontend/game assets,
+        messaging for build notifications, …) keeps it while coding.
+
+        Callers apply this only when the user hasn't pinned an explicit
+        selection (``--toolsets``, ``HERMES_TUI_TOOLSETS``, …); they never
+        override a pin. Returns the profile's toolset plus enabled MCP servers.
+        """
+        if self.config_mode != "focus":
+            return None
+        if self.profile.toolset is None:
+            return None
+        return [self.profile.toolset, *_enabled_mcp_servers(config)]
+
+    def system_blocks(self) -> list[str]:
+        """Stable system-prompt blocks for this posture (brief + workspace).
+
+        The operating brief carries a model-family edit-format nudge appended
+        to it (one cached string, not a separate block) so the model is steered
+        toward the `patch` mode it handles best — see ``_edit_format_line``.
+        """
+        if not self.is_coding:
+            return []
+        blocks: list[str] = []
+        if self.profile.guidance:
+            brief = self.profile.guidance
+            edit_line = _edit_format_line(self.model)
+            if edit_line:
+                brief = f"{brief}\n{edit_line}"
+            blocks.append(brief)
+        workspace = build_coding_workspace_block(self.cwd)
+        if workspace:
+            blocks.append(workspace)
+        return blocks
+
+    def compact_skill_categories(self) -> frozenset[str]:
+        """Skill categories to demote to names-only in the prompt's skill index.
+
+        Gated on the opt-in ``focus`` mode, like the toolset collapse: the
+        default posture leaves the skill index untouched. Users who didn't ask
+        for a lean prompt keep full entries for every category — index changes
+        under ``auto`` proved too surprising in practice, even names-only ones
+        (a demoted description is information the model no longer weighs when
+        deciding what to load).
+
+        Demoted — never hidden — even under ``focus``. An earlier revision
+        fully pruned these categories from the index, which caused silent
+        capability loss in a real workflow: agent-created skills are the
+        model's accumulated project memory (server-ops runbooks, learned
+        pitfalls, …), and models do not reliably reach for ``skills_list`` to
+        rediscover what the index stopped showing them. Names-only keeps every
+        skill loadable on recall while still cutting the description noise.
+        """
+        if not self.is_coding or self.config_mode != "focus":
+            return frozenset()
+        return frozenset(self.profile.compact_skill_categories)
+
+
+def resolve_runtime_mode(
+    *,
+    platform: Optional[str] = None,
+    cwd: Optional[str | Path] = None,
+    config: Optional[dict[str, Any]] = None,
+    model: Optional[str] = None,
+) -> RuntimeMode:
+    """Resolve the operating posture once. Cheap — a handful of ``stat`` calls.
+
+    This is the single entry point every domain should call. The returned
+    object is immutable and safe to cache for the session. Detection itself is
+    intentionally *not* memoized (see ``_detect_profile_name``) so a long-lived
+    process can't pin a stale posture; callers resolve once per session and
+    hold the result. ``model`` is recorded only to steer edit-format guidance;
+    it never affects detection.
+    """
+    resolved_cwd = _resolve_cwd(cwd)
+    mode = _coding_mode(config)
+    name = _detect_profile_name(
+        mode, (platform or "").strip().lower(), str(resolved_cwd)
+    )
+    return RuntimeMode(
+        profile=get_profile(name),
+        surface=platform or "",
+        cwd=resolved_cwd,
+        config_mode=mode,
+        model=model,
+    )
+
+
+# ── Back-compat surface (thin wrappers over RuntimeMode) ────────────────────
+
+
+def is_coding_context(
+    *,
+    platform: Optional[str] = None,
+    cwd: Optional[str | Path] = None,
+    config: Optional[dict[str, Any]] = None,
+) -> bool:
+    """Whether Hermes should operate in its coding posture right now."""
+    return resolve_runtime_mode(platform=platform, cwd=cwd, config=config).is_coding
+
+
+def coding_selection(
+    *,
+    platform: Optional[str] = None,
+    cwd: Optional[str | Path] = None,
+    config: Optional[dict[str, Any]] = None,
+) -> Optional[list[str]]:
+    """Toolset selection for the coding posture.
+
+    ``None`` unless the user opted into ``focus`` mode AND the posture is
+    active — the default coding posture never overrides configured toolsets.
+    """
+    return resolve_runtime_mode(
+        platform=platform, cwd=cwd, config=config
+    ).toolset_selection(config)
+
+
+def coding_system_blocks(
+    *,
+    platform: Optional[str] = None,
+    cwd: Optional[str | Path] = None,
+    config: Optional[dict[str, Any]] = None,
+    model: Optional[str] = None,
+) -> list[str]:
+    """Stable system-prompt blocks for the current posture (empty when general).
+
+    ``model`` steers the brief's edit-format nudge toward the model's family.
+    """
+    return resolve_runtime_mode(
+        platform=platform, cwd=cwd, config=config, model=model
+    ).system_blocks()
+
+
+def coding_compact_skill_categories(
+    *,
+    platform: Optional[str] = None,
+    cwd: Optional[str | Path] = None,
+    config: Optional[dict[str, Any]] = None,
+) -> frozenset[str]:
+    """Skill categories the active posture demotes to names-only in the index.
+
+    Empty outside the coding posture and outside the opt-in ``focus`` mode —
+    the default posture never touches the skill index. Under ``focus``,
+    demoted — never hidden: every skill name stays in the index and remains
+    loadable via ``skill_view`` / ``skills_list``; only descriptions are
+    dropped.
+    """
+    return resolve_runtime_mode(
+        platform=platform, cwd=cwd, config=config
+    ).compact_skill_categories()
+
+
+def _enabled_mcp_servers(config: Optional[dict[str, Any]]) -> list[str]:
+    """Names of MCP servers the user has enabled — kept in the coding posture.
+
+    MCP servers (figma, browser, tophat, …) are explicitly configured and part
+    of the coding workflow, not noise to strip.
+    """
+    try:
+        from hermes_cli.config import read_raw_config
+        from hermes_cli.tools_config import _parse_enabled_flag
+
+        servers = read_raw_config().get("mcp_servers") or {}
+        return [
+            str(name)
+            for name, cfg in servers.items()
+            if isinstance(cfg, dict)
+            and _parse_enabled_flag(cfg.get("enabled", True), default=True)
+        ]
+    except Exception:
+        return []
+
+
+# ── git/workspace probe ─────────────────────────────────────────────────────
+
+
+def _git(cwd: Path, *args: str) -> str:
+    try:
+        out = subprocess.run(
+            ["git", "-C", str(cwd), *args],
+            capture_output=True,
+            text=True,
+            timeout=_GIT_TIMEOUT,
+        )
+    except (OSError, subprocess.SubprocessError):
+        return ""
+    return out.stdout.strip() if out.returncode == 0 else ""
+
+
+def _parse_status(porcelain: str) -> tuple[dict[str, str], dict[str, int]]:
+    """Parse ``git status --porcelain=2 --branch`` into branch + counts."""
+    branch: dict[str, str] = {}
+    counts = {"staged": 0, "modified": 0, "untracked": 0, "conflicts": 0}
+    for line in porcelain.splitlines():
+        if line.startswith("# branch.head"):
+            branch["head"] = line.split(maxsplit=2)[-1]
+        elif line.startswith("# branch.upstream"):
+            branch["upstream"] = line.split(maxsplit=2)[-1]
+        elif line.startswith("# branch.ab"):
+            parts = line.split()
+            branch["ahead"], branch["behind"] = parts[2].lstrip("+"), parts[3].lstrip("-")
+        elif line.startswith(("1 ", "2 ")):
+            xy = line.split(maxsplit=2)[1]
+            if xy[0] != ".":
+                counts["staged"] += 1
+            if xy[1] != ".":
+                counts["modified"] += 1
+        elif line.startswith("u "):
+            counts["conflicts"] += 1
+        elif line.startswith("? "):
+            counts["untracked"] += 1
+    return branch, counts
+
+
+def _read_small(path: Path) -> str:
+    """Read a small text file, or ``""`` — never raises, never reads huge files."""
+    try:
+        if not path.is_file() or path.stat().st_size > _MAX_FACT_FILE_BYTES:
+            return ""
+        return path.read_text(encoding="utf-8", errors="replace")
+    except OSError:
+        return ""
+
+
+def _project_facts(root: Path) -> list[str]:
+    """Detected project facts for the workspace snapshot.
+
+    The point is to hand the model its *verify loop* up front — which manifest,
+    which package manager, and the exact test/lint/build commands — instead of
+    making it rediscover them every session. Cheap: stat calls plus reads of a
+    couple of small files; built once at prompt-build time (cache-safe).
+    """
+    facts: list[str] = []
+
+    manifests = [m for m in _PROJECT_MARKERS if m not in _CONTEXT_FILES and (root / m).is_file()]
+    package_managers = [
+        pm for lock, pm in (*_PY_LOCKFILES, *_JS_LOCKFILES) if (root / lock).is_file()
+    ]
+    if manifests:
+        line = f"- Project: {', '.join(manifests[:6])}"
+        if package_managers:
+            line += f" ({'/'.join(dict.fromkeys(package_managers))})"
+        facts.append(line)
+
+    verify: list[str] = []
+    if (root / "scripts" / "run_tests.sh").is_file():
+        verify.append("scripts/run_tests.sh")
+    if (root / "package.json").is_file():
+        try:
+            scripts = json.loads(_read_small(root / "package.json") or "{}").get("scripts") or {}
+        except (json.JSONDecodeError, AttributeError):
+            scripts = {}
+        js_pm = next((pm for lock, pm in _JS_LOCKFILES if (root / lock).is_file()), "npm")
+        verify.extend(f"{js_pm} run {name}" for name in _VERIFY_TARGETS if name in scripts)
+    if (root / "pytest.ini").is_file() or "[tool.pytest" in _read_small(root / "pyproject.toml"):
+        verify.append("pytest")
+    makefile = _read_small(root / "Makefile")
+    if makefile:
+        verify.extend(
+            f"make {name}" for name in _VERIFY_TARGETS
+            if re.search(rf"^{re.escape(name)}\s*:", makefile, re.MULTILINE)
+        )
+    if verify:
+        deduped = list(dict.fromkeys(verify))[:_MAX_VERIFY_COMMANDS]
+        facts.append(f"- Verify: {'; '.join(deduped)}")
+
+    context_files = [c for c in _CONTEXT_FILES if (root / c).is_file()]
+    if context_files:
+        facts.append(f"- Context files: {', '.join(context_files)}")
+
+    return facts
+
+
+def build_coding_workspace_block(cwd: Optional[str | Path] = None) -> str:
+    """Workspace snapshot for the system prompt (empty outside a workspace).
+
+    Git state (branch/status/commits) when the cwd is in a repo, plus detected
+    project facts (manifest, package manager, verify commands, context files)
+    — so marker-only (non-git) projects still get a snapshot.
+    """
+    resolved = _resolve_cwd(cwd)
+    git_root = _git_root(resolved)
+    root = git_root or _marker_root(resolved)
+    if root is None:
+        return ""
+
+    lines = ["Workspace (snapshot at session start — re-check with `git` before acting on it):"]
+    lines.append(f"- Root: {root}")
+
+    if git_root is not None:
+        branch, counts = _parse_status(_git(root, "status", "--porcelain=2", "--branch"))
+        head = branch.get("head", "")
+        if head and head != "(detached)":
+            line = f"- Branch: {head}"
+            if branch.get("upstream"):
+                line += f" \u2192 {branch['upstream']}"
+                ahead, behind = branch.get("ahead", "0"), branch.get("behind", "0")
+                if ahead != "0" or behind != "0":
+                    line += f" (ahead {ahead}, behind {behind})"
+            lines.append(line)
+        elif head == "(detached)":
+            lines.append("- Branch: (detached HEAD)")
+
+        # Linked worktree: the per-worktree git dir differs from the shared common dir.
+        # We surface the fact that it's a worktree (so the model knows branches/stashes
+        # are shared state) but deliberately do NOT expose the primary tree path —
+        # giving the model a second absolute path causes it to sometimes run commands
+        # in the wrong directory.
+        git_dir, common_dir = _git(root, "rev-parse", "--git-dir"), _git(root, "rev-parse", "--git-common-dir")
+        if git_dir and common_dir and Path(git_dir).resolve() != Path(common_dir).resolve():
+            lines.append("- Worktree: linked (git state shared with primary tree)")
+
+        dirty = [f"{n} {label}" for label, n in (
+            ("staged", counts["staged"]), ("modified", counts["modified"]),
+            ("untracked", counts["untracked"]), ("conflicts", counts["conflicts"]),
+        ) if n]
+        lines.append(f"- Status: {', '.join(dirty) if dirty else 'clean'}")
+
+        recent = _git(root, "log", "-3", "--pretty=%h %s")
+        if recent:
+            lines.append("- Recent commits:")
+            lines.extend(f"    {c}" for c in recent.splitlines())
+
+    lines.extend(_project_facts(root))
+    return "\n".join(lines)