prizmkit 1.1.101 → 1.1.104

package/bundled/VERSION.json

@@ -1,5 +1,5 @@
 {
-  "frameworkVersion": "1.1.101",
-  "bundledAt": "2026-07-06T10:58:41.729Z",
-  "bundledFrom": "fa258ae"
+  "frameworkVersion": "1.1.104",
+  "bundledAt": "2026-07-06T16:42:02.971Z",
+  "bundledFrom": "05c5c13"
 }

package/bundled/dev-pipeline/prizmkit_runtime/daemon.py

@@ -320,7 +320,7 @@ def _payload_command(family: RunnerFamily, paths, list_path: Path, options: Daem
 
 
 def _daemon_env(options: DaemonStartOptions) -> dict[str, str]:
-    env: dict[str, str] = {}
+    env: dict[str, str] = {"PRIZMKIT_LIVE_OUTPUT": "0"}
     for token in shlex.split(options.env_overrides):
         if "=" in token:
             key, value = token.split("=", 1)

package/bundled/dev-pipeline/prizmkit_runtime/runner_models.py

@@ -37,14 +37,16 @@ class RunnerEnvironment:
 
     stop_on_failure: bool = False
     auto_push: bool = False
-    use_worktree: bool = True
+    use_worktree: bool = False
     dev_branch: str = ""
     enable_deploy: bool = False
     verbose: bool = False
+    live_output: bool = True
     heartbeat_interval_seconds: int = 30
     heartbeat_stale_threshold_seconds: int = 600
     stale_kill_threshold_seconds: int = 600
     max_log_size: int | None = None
+    max_retries: int = 3
     strict_behavior_check: bool = True
 
     @classmethod
@@ -53,10 +55,11 @@ class RunnerEnvironment:
         return cls(
             stop_on_failure=_truthy(values.get("STOP_ON_FAILURE")),
             auto_push=_truthy(values.get("AUTO_PUSH")),
-            use_worktree=not _falsey(values.get("USE_WORKTREE")),
+            use_worktree=_truthy(values.get("USE_WORKTREE")),
             dev_branch=values.get("DEV_BRANCH", ""),
             enable_deploy=_truthy(values.get("ENABLE_DEPLOY")),
             verbose=_truthy(values.get("VERBOSE")),
+            live_output=not _falsey(values.get("PRIZMKIT_LIVE_OUTPUT")),
             heartbeat_interval_seconds=_int_value(values, "HEARTBEAT_INTERVAL", 30),
             heartbeat_stale_threshold_seconds=_int_value(
                 values,
@@ -65,6 +68,7 @@ class RunnerEnvironment:
             ),
             stale_kill_threshold_seconds=_int_value(values, "STALE_KILL_THRESHOLD", 600),
             max_log_size=_optional_int(values.get("MAX_LOG_SIZE")),
+            max_retries=max(0, _int_value(values, "MAX_RETRIES", 3)),
             strict_behavior_check=not _falsey(values.get("STRICT_BEHAVIOR_CHECK")),
         )
 
@@ -77,6 +81,7 @@ class RunnerInvocation:
     list_path: Path
     item_id: str = ""
     item_filter: str = ""
+    max_retries: int | None = None
     max_infra_retries: int = 3
     mode: str | None = None
     critic: bool | None = None
@@ -228,6 +233,7 @@ def parse_invocation(
     list_path = family.plan_path
     item_id = ""
     item_filter = ""
+    max_retries: int | None = None
     max_infra_retries = 3
     mode: str | None = None
     critic: bool | None = None
@@ -259,6 +265,11 @@ def parse_invocation(
             item_filter = args[index]
         elif arg.startswith("--features="):
             item_filter = arg.split("=", 1)[1]
+        elif arg == "--max-retries" and index + 1 < len(args):
+            index += 1
+            max_retries = max(0, _safe_int(args[index], 3))
+        elif arg.startswith("--max-retries="):
+            max_retries = max(0, _safe_int(arg.split("=", 1)[1], 3))
         elif arg == "--max-infra-retries" and index + 1 < len(args):
             index += 1
             max_infra_retries = _safe_int(args[index], 3)
@@ -293,6 +304,7 @@ def parse_invocation(
         list_path=list_path,
         item_id=item_id,
         item_filter=item_filter,
+        max_retries=max_retries,
         max_infra_retries=max_infra_retries,
         mode=mode,
         critic=critic,

package/bundled/dev-pipeline/prizmkit_runtime/runner_status.py

@@ -8,7 +8,7 @@ import sys
 from dataclasses import dataclass
 from pathlib import Path
 
-from .runner_models import RunnerFamily, RunnerInvocation
+from .runner_models import RunnerEnvironment, RunnerFamily, RunnerInvocation
 
 
 @dataclass(frozen=True)
@@ -30,6 +30,12 @@ class StatusCommandResult:
         return self.stdout.strip() or self.stderr.strip()
 
 
+def _resolved_max_retries(invocation: RunnerInvocation) -> int:
+    if invocation.max_retries is not None:
+        return max(0, invocation.max_retries)
+    return RunnerEnvironment.from_env().max_retries
+
+
 def run_status_action(
     family: RunnerFamily,
     action: str,
@@ -38,7 +44,7 @@ def run_status_action(
     item_id: str = "",
     session_status: str = "",
     session_id: str = "",
-    max_retries: int = 3,
+    max_retries: int | None = None,
     active_dev_branch: str = "",
     base_branch: str = "",
     last_fatal_error_code: str = "",
@@ -47,7 +53,7 @@ def run_status_action(
     progress_fingerprint: object | None = None,
     project_root: Path | None = None,
 ) -> StatusCommandResult:
-    """Invoke an existing updater script and parse JSON output when present."""
+    max_retries = _resolved_max_retries(invocation) if max_retries is None else max(0, max_retries)
     command = [
         sys.executable,
         str(family.updater_script),

package/bundled/dev-pipeline/prizmkit_runtime/runners.py

@@ -114,6 +114,8 @@ def _run_pipeline(family: RunnerFamily, invocation: RunnerInvocation, paths) ->
         details.append(f"{item_id}:{final_status}")
         if final_status == "completed":
             continue
+        if final_status == "pending":
+            continue
         if RunnerEnvironment.from_env().stop_on_failure:
             return PipelineRunResult(False, processed, "stop_on_failure", final_status, tuple(details))
         if final_status in {"failed", "needs_info"}:
@@ -218,6 +220,7 @@ def _process_item(
                 heartbeat_path=session_paths.heartbeat_json,
                 effort=config.effort,
                 verbose=env.verbose,
+                live_output=env.live_output,
                 use_stream_json=stream_json,
                 stale_kill_threshold_seconds=env.stale_kill_threshold_seconds,
             )
@@ -401,6 +404,7 @@ def _resolve_invocation_paths(invocation: RunnerInvocation, project_root: Path)
         list_path=(project_root / invocation.list_path).resolve(),
         item_id=invocation.item_id,
         item_filter=invocation.item_filter,
+        max_retries=invocation.max_retries,
         max_infra_retries=invocation.max_infra_retries,
         mode=invocation.mode,
         critic=invocation.critic,
@@ -418,5 +422,5 @@ def _help_text(family: RunnerFamily) -> str:
     return (
         f"Python foreground runner for {family.kind}.\n"
         "Supported legacy options include plan list path, item id, --features for feature, "
-        "--max-infra-retries, --mode, --critic/--no-critic, --resume-phase, and --dry-run."
+        "--max-retries, --max-infra-retries, --mode, --critic/--no-critic, --resume-phase, and --dry-run."
     )

package/bundled/dev-pipeline/prizmkit_runtime/sessions.py

@@ -8,6 +8,7 @@ import os
 import queue
 import shutil
 import subprocess
+import sys
 import threading
 import time
 from collections import Counter
@@ -45,6 +46,7 @@ class AISessionConfig:
     heartbeat_path: Path | None = None
     effort: str | None = None
     verbose: bool = False
+    live_output: bool = False
     use_stream_json: bool = False
     stream_json_format: str = ""
     permission_mode: str = "dangerously-skip-permissions"
@@ -152,7 +154,14 @@ def build_ai_session_command(config: AISessionConfig) -> AISessionCommand:
     platform = _normalize_platform(config.platform)
     prompt = _prompt_text(config.prompt_path)
     if platform == "claude":
-        argv = [config.cli_command, "-p", prompt, "--dangerously-skip-permissions"]
+        argv = [
+            config.cli_command,
+            "-p",
+            prompt,
+            "--dangerously-skip-permissions",
+            "--disallowedTools",
+            "EnterWorktree",
+        ]
         if config.verbose or config.use_stream_json:
             argv.append("--verbose")
         if config.use_stream_json:
@@ -175,6 +184,7 @@ def build_ai_session_command(config: AISessionConfig) -> AISessionCommand:
         timeout = _codex_subagent_timeout(config)
         if timeout > 0:
             argv.extend(["--config", f"agents.job_max_runtime_seconds={timeout}"])
+        argv.extend(["--profile", "no-worktree"])
         argv.extend(["exec", "--cd", str(config.cwd), "--skip-git-repo-check"])
         if config.use_stream_json:
             argv.append("--json")
@@ -186,7 +196,7 @@ def build_ai_session_command(config: AISessionConfig) -> AISessionCommand:
         argv.append("-")
         return AISessionCommand(tuple(argv), prompt, "codex", "stdin")
 
-    argv = [config.cli_command, "--print", "-y"]
+    argv = [config.cli_command, "--print", "-y", "--disallowedTools", "EnterWorktree"]
     if config.verbose:
         argv.append("--verbose")
     if config.use_stream_json:
@@ -355,12 +365,45 @@ def _write_progress_summary(progress_path: Path | None, summary: ProgressSummary
     progress_path.write_text(json.dumps(summary.to_dict(), indent=2, sort_keys=True) + "\n", encoding="utf-8")
 
 
+def _emit_live_line(message: str) -> None:
+    """Print foreground runner status without interfering with file logging."""
+    print(message, file=sys.stderr, flush=True)
+
+
+def _emit_live_output(line: str) -> None:
+    """Forward AI CLI output for verbose foreground runs."""
+    sys.stdout.write(line)
+    sys.stdout.flush()
+
+
+def _emit_session_start(config: AISessionConfig, command: AISessionCommand, session_log: Path, backup_log: Path) -> None:
+    _emit_live_line("[prizmkit] AI session started")
+    _emit_live_line(f"[prizmkit] Provider: {command.provider}")
+    _emit_live_line(f"[prizmkit] Prompt: {config.prompt_path}")
+    _emit_live_line(f"[prizmkit] Session log: {session_log}")
+    _emit_live_line(f"[prizmkit] Backup log: {backup_log}")
+    if config.progress_path is not None:
+        _emit_live_line(f"[prizmkit] Progress: {config.progress_path}")
+    if config.heartbeat_path is not None:
+        _emit_live_line(f"[prizmkit] Heartbeat: {config.heartbeat_path}")
+    if config.verbose:
+        _emit_live_line("[prizmkit] VERBOSE=1: forwarding live AI CLI output below")
+
+
+def _emit_session_finish(result: AISessionResult) -> None:
+    duration = max(0.0, result.ended_epoch - result.started_epoch)
+    status = result.termination_reason or f"exit_code={result.exit_code}"
+    _emit_live_line(f"[prizmkit] AI session finished ({status}, {duration:.1f}s)")
+    _emit_live_line(f"[prizmkit] Session log: {result.session_log}")
+
+
 def _reader_thread(
     pipe,
     session_log: Path,
     backup_log: Path,
     output_queue: "queue.Queue[str]",
     progress_path: Path | None = None,
+    live_output: bool = False,
 ) -> None:
     lines: list[str] = []
     tracker = None
@@ -379,6 +422,8 @@ def _reader_thread(
             with backup_log.open("a", encoding="utf-8") as backup_fh:
                 backup_fh.write(line)
             output_queue.put(line)
+            if live_output:
+                _emit_live_output(line)
             lines.append(line)
             if progress_path is not None:
                 if tracker is not None:
@@ -464,12 +509,22 @@ class AISessionLauncher:
             environment_unsets=("CLAUDECODE",),
             stdin_text=command.stdin_text,
         )
+        if self.config.live_output:
+            _emit_session_start(self.config, command, session_log, backup_log)
+
         handle = launch_process(spec, stderr=subprocess.STDOUT)
         assert handle.popen is not None
         output_queue: queue.Queue[str] = queue.Queue()
         reader = threading.Thread(
             target=_reader_thread,
-            args=(handle.popen.stdout, session_log, backup_log, output_queue, self.config.progress_path),
+            args=(
+                handle.popen.stdout,
+                session_log,
+                backup_log,
+                output_queue,
+                self.config.progress_path,
+                self.config.live_output and self.config.verbose,
+            ),
             daemon=True,
         )
         reader.start()
@@ -536,7 +591,7 @@ class AISessionLauncher:
             self.config.progress_path.write_text(json.dumps(merged_progress, indent=2, sort_keys=True) + "\n", encoding="utf-8")
         final_summary = _summary_from_progress_dict(merged_progress, summary)
         ended = time.time()
-        return AISessionResult(
+        result = AISessionResult(
             command=command,
             exit_code=handle.popen.returncode,
             timed_out=timed_out,
@@ -550,3 +605,6 @@ class AISessionLauncher:
             started_epoch=started,
             ended_epoch=ended,
         )
+        if self.config.live_output:
+            _emit_session_finish(result)
+        return result

package/bundled/dev-pipeline/scripts/generate-bootstrap-prompt.py

@@ -676,7 +676,7 @@ def determine_pipeline_mode(complexity):
 SECTION_TO_SKILL = {
     "phase0-init": ("prizmkit-init", "Project Bootstrap",
                     [".prizmkit/prizm-docs/root.prizm", ".prizmkit/config.json"]),
-    "phase0-test-baseline": ("test-baseline", "Test Baseline", []),
+    "phase0-test-baseline": ("test-baseline", "Detect Test Commands", []),
     "phase-context-snapshot": ("context-snapshot", "Build Context Snapshot",
                                [".prizmkit/specs/{slug}/context-snapshot.md"]),
     "phase-specify-plan": ("context-snapshot-and-plan", "Specify & Plan",
@@ -1559,6 +1559,18 @@ def assemble_sections(pipeline_mode, sections_dir, init_done, is_resume,
                           load_section(sections_dir,
                                        "phase-implement-agent.md")))
 
+    # Verification Gates are included in Task Contract. Keep AC in one place so
+    # background context and implementation prompts cannot redefine scope.
+    # --- Review ---
+    if pipeline_mode == "full":
+        sections.append(("phase-review",
+                          load_section(sections_dir,
+                                       "phase-review-full.md")))
+    else:
+        sections.append(("phase-review",
+                          load_section(sections_dir,
+                                       "phase-review-agent.md")))
+
     # --- Test Failure Recovery Protocol (tier-specific) ---
     if pipeline_mode == "lite":
         sections.append(("test-failure-recovery",
@@ -1574,18 +1586,6 @@ def assemble_sections(pipeline_mode, sections_dir, init_done, is_resume,
                       load_section(sections_dir,
                                    "phase-prizmkit-test.md")))
 
-    # Verification Gates are included in Task Contract. Keep AC in one place so
-    # background context and implementation prompts cannot redefine scope.
-    # --- Review (only for agent tiers) ---
-    if pipeline_mode == "full":
-        sections.append(("phase-review",
-                          load_section(sections_dir,
-                                       "phase-review-full.md")))
-    elif pipeline_mode == "standard":
-        sections.append(("phase-review",
-                          load_section(sections_dir,
-                                       "phase-review-agent.md")))
-
     # --- Browser Verification (conditional, tool-aware) ---
     if browser_enabled:
         if browser_tool == "opencli":

package/bundled/dev-pipeline/templates/agent-prompts/dev-implement.md

@@ -61,7 +61,7 @@ Before returning, append `## Implementation Log` to `context-snapshot.md` with:
 - Do not run git commands; staging and commit are handled by the orchestrator.
 - **Edit safety**: If an Edit fails with 'String to replace not found', grep for the target text before retrying. Never guess file offsets — verify them with a Read or grep first.
 - **Read safety**: If 3 consecutive Reads to the same file return 'shorter than offset' or 'Wasted call', STOP and report BLOCKED.
-- **Test early**: Run `{{TEST_CMD}}` after every 3 successful Edit operations. Capture output to /tmp/test-out.txt and grep for failures.
+- **Test gate placement**: Do not run mandatory periodic or full-suite tests here. The scoped/full test gate runs after code review; only use tiny targeted checks when needed for local debugging.
 
 Do not return success unless:
 1. implementation tasks are complete;

package/bundled/dev-pipeline/templates/agent-prompts/reviewer-review.md

@@ -2,6 +2,6 @@
 1. Read `.prizmkit/specs/{{FEATURE_SLUG}}/spec.md` (if it exists) for goals and Verification Gates; if spec.md does not exist, read `.prizmkit/specs/{{FEATURE_SLUG}}/context-snapshot.md` Section 1 Task Contract instead
 2. Read `.prizmkit/specs/{{FEATURE_SLUG}}/plan.md` for architecture decisions and completed tasks
 3. Run /prizmkit-code-review with artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/. The skill will run its internal review-fix loop (Reviewer → filter → Dev fix, max 3 rounds) and write review-report.md.
-4. Run the full test suite using `{{TEST_CMD}}`. When running tests: `({{TEST_CMD}}) 2>&1 | tee /tmp/review-test-out.txt | tail -20`, then grep `/tmp/review-test-out.txt` for details — do NOT re-run the suite multiple times.
+4. Do not run test suites here. The orchestrator runs `/prizmkit-test` after review completes and then fixes code/tests until the scoped gate passes.
 5. review-report.md will be written to .prizmkit/specs/{{FEATURE_SLUG}}/ by prizmkit-code-review.
 Report: verdict (PASS/NEEDS_FIXES), number of rounds, findings fixed/rejected."

package/bundled/dev-pipeline/templates/bootstrap-prompt.md

@@ -30,9 +30,9 @@ Infer what needs to be done from the feature context above and follow the standa
 
 3. **Implement**: Run `/prizmkit-implement` with `artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` to execute the plan using TDD (write tests first, then implement).
 
-4. **Test**: Run `/prizmkit-test scope=this-change artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` to generate and verify tests only for this feature's changed scope. Do not use bugfix/refactor artifact directories for this gate.
+4. **Review**: Run `/prizmkit-code-review` with `artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` to review and auto-fix changes against the spec (internal review-fix loop, max 3 rounds).
 
-5. **Review**: Run `/prizmkit-code-review` with `artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` to review and auto-fix changes against the spec (internal review-fix loop, max 3 rounds).
+5. **Test**: Run `/prizmkit-test scope=this-change artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` to generate and verify tests only for this feature's changed scope after review. Do not use bugfix/refactor artifact directories for this gate.
 
 6. **Retrospective**: Run `/prizmkit-retrospective` to sync `.prizmkit/prizm-docs/` with code changes.
 

package/bundled/dev-pipeline/templates/bootstrap-tier1.md

@@ -145,9 +145,9 @@ You know this project's tech stack. Identify ALL test commands that apply (e.g.,
 **3b.** Run `/prizmkit-implement` — this handles the full implementation cycle:
 - Reads plan.md Tasks section from `.prizmkit/specs/{{FEATURE_SLUG}}/`
 - Reads context from `context-snapshot.md` (Prizm docs, TRAPS, file manifest)
-- Implements task-by-task with TDD, marking each `[x]` immediately
+- Implements task-by-task, marking each `[x]` immediately
 - Creates/updates L2 `.prizm` docs when creating new modules or significantly modifying existing ones — AI selectively decides which modules warrant L2 based on complexity and importance
-- Runs tests using `TEST_CMD` after each task
+- Defers scoped/full test execution until after review
 - Writes '## Implementation Log' to `context-snapshot.md`
 
 **3b-checkpoint.** Context management — if plan.md has more than 5 tasks, update durable checkpoints/artifacts after every 3 tasks, minimize output, and continue. If context overflow occurs in a headless run, the runner starts a fresh continuation session from those checkpoints and artifacts. In an interactive Claude Code session operated manually outside the headless pipeline, `/compact` may be used as an optional convenience only.
@@ -155,10 +155,9 @@ You know this project's tech stack. Identify ALL test commands that apply (e.g.,
 **3c.** After implement completes, verify:
 1. All tasks in plan.md are `[x]`
 2. **Verification Gate Self-Check**: for each acceptance criterion, write one line of evidence (test name / file:line / fixture path). Any gate without evidence is BLOCKED, not success.
-3. **Layered test strategy**: debug with single-file tests (`npx vitest run <file> -t '<name>'`), only run full suite as gate when single-file is green. Never run full suite while a single-file test is still red.
-4. Run the full test suite to ensure nothing is broken
-5. Verify each acceptance criterion from Section 1 of context-snapshot.md is met — check mentally, do NOT re-read files you already wrote
-6. If any criterion is not met, fix it now using the convergence-based recovery loop below
+3. **Layered test strategy**: do not run mandatory periodic/full-suite tests during implementation. Use only tiny targeted checks when needed for local debugging; the scoped feature test gate runs after review.
+4. Verify each acceptance criterion from Section 1 of context-snapshot.md is met — check mentally, do NOT re-read files you already wrote
+5. If any criterion is not met, fix it now using the convergence-based recovery loop below
 
 **CP-2**: All acceptance criteria met, all tests pass.
 
@@ -177,7 +176,7 @@ When tests fail, use convergence recovery — keep fixing while progress is bein
 
 ### Scoped Feature Test Gate — PrizmKit Test
 
-Run `/prizmkit-test scope=this-change artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` after implementation and before review/commit.
+Run `/prizmkit-test scope=this-change artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` after review and before commit.
 
 Before invoking the skill, create `.prizmkit/specs/{{FEATURE_SLUG}}/.prizmkit-test-started` so stale reports are ignored.
 

package/bundled/dev-pipeline/templates/bootstrap-tier2.md

@@ -302,7 +302,7 @@ If GATE:MISSING — send message to Dev (re-spawn if needed): "Write the '## Imp
 
 ### Scoped Feature Test Gate — PrizmKit Test
 
-Run `/prizmkit-test scope=this-change artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` after implementation and before review/commit.
+Run `/prizmkit-test scope=this-change artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` after review and before commit.
 
 Before invoking the skill, create `.prizmkit/specs/{{FEATURE_SLUG}}/.prizmkit-test-started` so stale reports are ignored.
 

package/bundled/dev-pipeline/templates/bootstrap-tier3.md

@@ -320,7 +320,7 @@ All tasks `[x]`, tests pass.
 
 ### Scoped Feature Test Gate — PrizmKit Test
 
-Run `/prizmkit-test scope=this-change artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` after implementation and before review/commit.
+Run `/prizmkit-test scope=this-change artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` after review and before commit.
 
 Before invoking the skill, create `.prizmkit/specs/{{FEATURE_SLUG}}/.prizmkit-test-started` so stale reports are ignored.
 

package/bundled/dev-pipeline/templates/sections/phase-implement-agent.md

@@ -19,15 +19,15 @@ grep -c '^\- \[ \]' .prizmkit/specs/{{FEATURE_SLUG}}/plan.md 2>/dev/null || true
 **3b.** Run `/prizmkit-implement` directly — you (the orchestrator) execute the full cycle:
 - Read plan.md Tasks from `.prizmkit/specs/{{FEATURE_SLUG}}/`
 - Use context-snapshot.md Section 4 File Manifest for targeted reads — do NOT re-read files already summarized there
-- Implement task-by-task with TDD, marking each `[x]` immediately
+- Implements task-by-task, marking each `[x]` immediately
 - Creates/updates L2 `.prizm` docs when creating new modules
-- Runs tests using `TEST_CMD`
+- Defers scoped/full test execution until after the code-review loop completes
 - If plan.md has >5 tasks: update checkpoints after every 3 tasks
 
-**3c. Layered test strategy** (replaces "run full suite every 3 Edits"):
-- Debugging a single test: `npx vitest run <test-file> -t '<test name>'` (or per-stack equivalent). Capture to `/tmp/test-out.txt`, grep for failures.
-- Only when single-file is green, run ONE full `{{TEST_CMD}}` as the gate.
-- Never run full suite while a single-file test is still red.
+**3c. Focused checks only (no test gate here)**:
+- During implementation, do not run mandatory periodic or full-suite tests.
+- If a tiny targeted check is needed to understand a local failure, run only the smallest relevant command and capture output to `/tmp/test-out.txt`.
+- The scoped feature test gate runs after code review; do not declare implementation success based on an early full-suite run.
 
 **3d. Verification Gate Self-Check (BLOCKING — before declaring implement done)**:
 For each Verification Gate from the Task Contract, write one evidence line to Implementation Log:
@@ -45,7 +45,7 @@ For each Verification Gate from the Task Contract, write one evidence line to Im
 - Gate Evidence section
 - unresolved blockers
 
-**CP-2**: All tasks `[x]`, Implementation Log written with Gate Evidence, Test Failure Recovery Protocol's Success Rule satisfied. Blocked gates → `failure-log.md`, not success.
+**CP-2**: All tasks `[x]`, Implementation Log written with Gate Evidence, and blocked gates documented in `failure-log.md`. Test execution is deferred until after code review.
 
 **Checkpoint update**:
 ```bash

package/bundled/dev-pipeline/templates/sections/phase-implement-full.md

@@ -20,15 +20,15 @@ Save pre-existing failing tests as `BASELINE_FAILURES`.
 **3b.** Run `/prizmkit-implement` directly — you (the orchestrator) execute the full cycle:
 - Read plan.md Tasks from `.prizmkit/specs/{{FEATURE_SLUG}}/`
 - Use context-snapshot.md Section 4 File Manifest for targeted reads — do NOT re-read files already summarized there
-- Implement task-by-task with TDD, marking each `[x]` immediately
+- Implements task-by-task, marking each `[x]` immediately
 - Creates/updates L2 `.prizm` docs when creating new modules
-- Runs tests using `TEST_CMD`
+- Defers scoped/full test execution until after the code-review loop completes
 - If plan.md has >5 tasks: update checkpoints after every 3 tasks
 
-**3c. Layered test strategy** (replaces "run full suite every 3 Edits"):
-- Debugging a single test: `npx vitest run <test-file> -t '<test name>'` (or per-stack equivalent). Capture to `/tmp/test-out.txt`, grep for failures.
-- Only when single-file is green, run ONE full `{{TEST_CMD}}` as the gate.
-- Never run full suite while a single-file test is still red.
+**3c. Focused checks only (no test gate here)**:
+- During implementation, do not run mandatory periodic or full-suite tests.
+- If a tiny targeted check is needed to understand a local failure, run only the smallest relevant command and capture output to `/tmp/test-out.txt`.
+- The scoped feature test gate runs after code review; do not declare implementation success based on an early full-suite run.
 
 **3d. Verification Gate Self-Check (BLOCKING — before declaring implement done)**:
 For each Verification Gate from the Task Contract, write one evidence line to Implementation Log:
@@ -46,7 +46,7 @@ For each Verification Gate from the Task Contract, write one evidence line to Im
 - Gate Evidence section
 - unresolved blockers
 
-**CP-2**: All tasks `[x]`, Implementation Log written with Gate Evidence, Test Failure Recovery Protocol's Success Rule satisfied. Blocked gates → `failure-log.md`, not success.
+**CP-2**: All tasks `[x]`, Implementation Log written with Gate Evidence, and blocked gates documented in `failure-log.md`. Test execution is deferred until after code review.
 
 **Checkpoint update**:
 ```bash

package/bundled/dev-pipeline/templates/sections/phase-implement-lite.md

@@ -1,33 +1,24 @@
-### Implement + Test
+### Implement
 
 **Protocol references**:
 - Follow Context Budget Rules §8 for scaffold/generated files.
 - Follow Context Budget Rules §9 before package install/build commands that resolve dependencies.
 - Follow Context Budget Rules §10 after build/compile commands.
 
-**3a.** Detect test commands and record baseline:
-
-You know this project's tech stack. Identify ALL test commands that apply (e.g., `go test ./...`, `npm test`, `cargo test`, `pytest`, `make test`, etc.). Record them as `TEST_CMDS`. Then record baseline:
-```bash
-# Run each test command, capture output
-($TEST_CMD) 2>&1 | tee /tmp/test-baseline.txt | tail -20
-```
-
-**3b.** Run `/prizmkit-implement` — this handles the full implementation cycle:
+**3a.** Run `/prizmkit-implement` — this handles the implementation cycle:
 - Reads plan.md Tasks section from `.prizmkit/specs/{{FEATURE_SLUG}}/`
 - Reads context from `context-snapshot.md` (Prizm docs, TRAPS, file manifest)
-- Implements task-by-task with TDD, marking each `[x]` immediately
+- Implements task-by-task, marking each `[x]` immediately
 - Creates/updates L2 `.prizm` docs when creating new modules or significantly modifying existing ones — AI selectively decides which modules warrant L2 based on complexity and importance
-- Runs tests using `TEST_CMD` after each task
+- Does not run the scoped/full test gate; testing happens after code review in the PrizmKit Test phase
 - Writes '## Implementation Log' to `context-snapshot.md`
 
-**3c.** After implement completes, verify:
+**3b.** After implement completes, verify:
 1. All tasks in plan.md are `[x]`
-2. Run the full test suite to ensure no new regressions remain
-3. Verify each Verification Gate from the Task Contract — check mentally, do NOT re-read files you already wrote
-4. If any gate is unmet or blocked, follow the Test Failure Recovery Protocol
+2. Verify each Verification Gate from the Task Contract using code/file evidence already in context — do NOT re-read files you already wrote
+3. If any gate is unmet or blocked, write `failure-log.md` and stop for recovery
 
-**CP-2**: Implementation may proceed only when all tasks are `[x]` and the Test Failure Recovery Protocol's Success Rule is satisfied. Blocked gates must be documented in `failure-log.md` and are not success.
+**CP-2**: Implementation may proceed only when all tasks are `[x]` and blocked gates are documented in `failure-log.md`. Test execution is deferred until after code review.
 
 
 **Checkpoint update**: Run the update script to set step `prizmkit-implement` to `"completed"`:

package/bundled/dev-pipeline/templates/sections/phase-prizmkit-test.md

@@ -1,6 +1,6 @@
 ### Scoped Feature Test Gate — PrizmKit Test
 
-**Goal**: Generate and verify tests for this feature's changed scope before code review without embedding the full gate implementation in this prompt.
+**Goal**: After code review completes, generate and verify tests for this feature's changed scope without embedding the full gate implementation in this prompt.
 
 Create a start marker immediately before invoking the skill so the gate can reject stale reports from older sessions:
 
@@ -34,6 +34,6 @@ python3 "$PIPELINE_DIR/scripts/prizmkit-test-gate.py" check \
 
 Gate outcomes:
 - `GATE:PASS` → append the report path and a 3-5 bullet summary to `context-snapshot.md` under `## PrizmKit Test Gate`, then proceed to the next checkpoint step. The gate script already wrote `test-report-path.txt` and marked `prizmkit-test` completed.
-- `GATE:FAIL` with `NEEDS_FIXES` → read the report's `In-Scope Failures`, fix the feature implementation or generated tests within this feature scope, rerun `/prizmkit-test`, and repeat the gate check. Do not proceed to code review while the verdict is not `PASS`.
+- `GATE:FAIL` with `NEEDS_FIXES` → read the report's `In-Scope Failures`, fix the feature implementation or generated tests within this feature scope, rerun `/prizmkit-test`, and repeat the gate check. Do not proceed to retrospective/commit while the verdict is not `PASS`.
 - `GATE:FAIL` with `BLOCKED`, wrong mode, wrong artifact dir, unreadable report, stale scope, boundary failure, validator failure, or checkpoint update failure → write `.prizmkit/specs/{{FEATURE_SLUG}}/failure-log.md` with the scoped test failure and stop for recovery.
 - `GATE:MISSING` → perform one bounded status check: inspect `/prizmkit-test` output and `.prizmkit/test/` for a report directory. If no current-run report exists, write `failure-log.md` and stop for recovery.

package/bundled/dev-pipeline/templates/sections/phase-review-agent.md

@@ -1,8 +1,6 @@
 ### Review — Code Review
 
-Precondition: the `prizmkit-test` checkpoint is completed and `.prizmkit/specs/{{FEATURE_SLUG}}/test-report-path.txt` points to the scoped test report.
-
-Before invoking review, read `.prizmkit/specs/{{FEATURE_SLUG}}/test-report-path.txt`, read the referenced `/prizmkit-test` report, and confirm the summary in `context-snapshot.md` under `## PrizmKit Test Gate` matches it. The code-review skill must receive this context through `artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` and the Reviewer must assess generated/updated test quality, `In-Scope Failures`, `Baseline Failures`, `Out of Scope Gaps`, boundary coverage if present, and the scoped report verdict.
+Review the completed implementation before the scoped feature test gate runs. Use `artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` so the code-review skill can read spec, plan, context snapshot, implementation log, and current diff artifacts. Do not require or read a prior `/prizmkit-test` report here; test generation and test-fix looping happen in the next phase.
 
 Run `/prizmkit-code-review` with `artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/`.
 

package/bundled/dev-pipeline/templates/sections/phase-review-full.md

@@ -1,8 +1,6 @@
 ### Review — Code Review
 
-Precondition: the `prizmkit-test` checkpoint is completed and `.prizmkit/specs/{{FEATURE_SLUG}}/test-report-path.txt` points to the scoped test report.
-
-Before invoking review, read `.prizmkit/specs/{{FEATURE_SLUG}}/test-report-path.txt`, read the referenced `/prizmkit-test` report, and confirm the summary in `context-snapshot.md` under `## PrizmKit Test Gate` matches it. The code-review skill must receive this context through `artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` and the Reviewer must assess generated/updated test quality, `In-Scope Failures`, `Baseline Failures`, `Out of Scope Gaps`, boundary coverage if present, and the scoped report verdict.
+Review the completed implementation before the scoped feature test gate runs. Use `artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` so the code-review skill can read spec, plan, context snapshot, implementation log, and current diff artifacts. Do not require or read a prior `/prizmkit-test` report here; test generation and test-fix looping happen in the next phase.
 
 Run `/prizmkit-code-review` with `artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/`.
 

package/bundled/dev-pipeline/templates/sections/phase0-test-baseline.md

@@ -1,15 +1,10 @@
-### Phase 0: Record Test Baseline & Detect Test Commands
+### Phase 0: Detect Test Commands
 
 **Step 1 — Detect test commands**: You know this project's tech stack. Identify ALL test commands that apply (e.g., `go test ./...`, `npm test`, `cargo test`, `pytest`, `make test`, etc.). Record them as `TEST_CMDS`.
 
-**Step 2 — Record pre-existing failure baseline**:
-```bash
-# Run each test command, capture output
-($TEST_CMD) 2>&1 | tee /tmp/test-baseline.txt | tail -20
-```
-Save the list of **pre-existing failing tests** (if any) as `BASELINE_FAILURES`. These are known failures that existed before this session — Dev must NOT be blamed for them, but must list them in COMPLETION_SIGNAL.
+**Step 2 — Do not run tests yet**: This phase only records the commands that will be used by the post-review PrizmKit Test gate. Do not run baseline, periodic, or full-suite tests here; test execution happens after code review.
 
-> **Test Output Rule**: Always capture test output to a temp file (`tee /tmp/test-out.txt`). Then grep the file instead of re-running the suite.
+> **Test Output Rule**: When the post-review gate runs, capture test output to a temp file (`tee /tmp/test-out.txt`). Then grep the file instead of re-running the suite.
 
 
 **Checkpoint update**: Run the update script to set step `test-baseline` to `"completed"`: