prizmkit 1.1.136 → 1.1.138

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

@@ -1,23 +1,23 @@
-### Implement — Orchestrator Direct (no Dev subagent)
+### Implement — Plan Execution
 
-**Rationale**: Development is deterministic plan execution — the orchestrator already built full context in Phase 1. Spawning a Dev subagent forces it to rebuild context (re-reading files the orchestrator already read), wastes context, and risks worktree double-edit. The orchestrator implements directly.
+Execute the reviewed plan through the active implementation skill. The skill and runtime determine the execution topology; preserve the shared artifact directory, active checkout, and task checkpoints.
 
 Before starting, check plan.md Tasks:
 ```bash
 {{RUNTIME_HELPER_CMD}} text count .prizmkit/specs/{{FEATURE_SLUG}}/plan.md --pattern "- [ ]"
 ```
 - If result is `0` (all tasks already `[x]`) → **SKIP to Review**.
-- If non-zero → implement directly below.
+- If non-zero → execute the remaining tasks below.
 
 **Build artifacts rule**: After any build/compile command, ensure output is in `.gitignore`. Never commit binaries/build output.
 
 **3a.** Run `/prizmkit-implement` directly:
 - Continue from the `spec.md` / `plan.md` / `context-snapshot.md` content you created in Phase 1-2; do not re-read self-authored artifacts just because implementation starts
 - Re-read `plan.md` or `context-snapshot.md` only when resuming/continuing, when current context was compacted or lost, or when task checkbox state may have changed outside this session
-- Implementation remains in the current Main Agent; do not delegate it to a Dev subagent
+- Follow the implementation skill's current execution and delegation contract
 - Use context-snapshot.md Section 4 File Manifest for targeted source reads; avoid re-reading source files already summarized there unless they changed after the snapshot
 - Implements task-by-task, marking each `[x]` immediately
-- Remeber to update checkpoints after finishing a series of tasks
+- Remember to update checkpoints after finishing a series of tasks
 
 **3b. Focused checks only (no test gate here)**:
 - During implementation, do not run mandatory periodic or full-suite tests.

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

@@ -1,23 +1,23 @@
-### Implement — Orchestrator Direct (no Dev subagent)
+### Implement — Plan Execution
 
-**Rationale**: Development is deterministic plan execution — the orchestrator already built full context in Phase 1-2. Spawning a Dev subagent forces it to rebuild context (re-reading files the orchestrator already read), wastes context, and risks worktree double-edit. The orchestrator implements directly.
+Execute the reviewed plan through the active implementation skill. The skill and runtime determine the execution topology; preserve the shared artifact directory, active checkout, and task checkpoints.
 
 Before starting, check plan.md Tasks:
 ```bash
 {{RUNTIME_HELPER_CMD}} text count .prizmkit/specs/{{FEATURE_SLUG}}/plan.md --pattern "- [ ]"
 ```
 - If result is `0` (all tasks already `[x]`) → **SKIP to Review**.
-- If non-zero → implement directly below.
+- If non-zero → execute the remaining tasks below.
 
 **Build artifacts rule**: After any build/compile command, ensure output is in `.gitignore`. Never commit binaries/build output.
 
 **3a.** Run `/prizmkit-implement` directly:
 - Continue from the `spec.md` / `plan.md` / `context-snapshot.md` content you created in Phase 1-2; do not re-read self-authored artifacts just because implementation starts
 - Re-read `plan.md` or `context-snapshot.md` only when resuming/continuing, when current context was compacted or lost, or when task checkbox state may have changed outside this session
-- Implementation remains in the current Main Agent; do not delegate it to a Dev subagent
+- Follow the implementation skill's current execution and delegation contract
 - Use context-snapshot.md Section 4 File Manifest for targeted source reads; avoid re-reading source files already summarized there unless they changed after the snapshot
 - Implements task-by-task, marking each `[x]` immediately
-- Remeber to update checkpoints after finishing a series of tasks
+- Remember to update checkpoints after finishing a series of tasks
 
 **3b. Focused checks only (no test gate here)**:
 - During implementation, do not run mandatory periodic or full-suite tests.

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

@@ -1,18 +1,8 @@
 ### Review — Code Review
 
-Review the completed implementation before the scoped feature test gate runs. Use `artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` so `/prizmkit-code-review` can read the spec, plan, context snapshot, implementation log, and current diff. Do not require a prior `/prizmkit-test` report here; the next phase owns the formal test gate.
+Review the completed implementation before the scoped feature test gate runs. Use `artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` so `/prizmkit-code-review` can read the spec, plan, context snapshot, implementation log, and current diff. Follow the skill's current review contract for review execution, repairs, verification, evidence, and handoff. Do not invoke another review skill or add a second review path outside that contract.
 
-Run `/prizmkit-code-review` with `artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/`. The skill runs a mandatory Main-Agent review loop for up to ten completed rounds: the Main Agent reviews the complete current change, rejects unsupported candidate findings, directly repairs accepted findings, verifies repairs, and repeats until `accepted = 0` with no unresolved findings. After convergence, one skill-owned independent Reviewer is optional only when the strict structural capability gate passes; otherwise no Reviewer is created.
-
-The Code Review execution boundary is closed for this phase:
-
-- Do not delegate the mandatory Main-Agent review or invoke another review skill or workflow.
-- Do not add review execution outside the single optional Reviewer owned and gated by `prizmkit-code-review`.
-- The optional Reviewer must be structurally read-only, unable to execute arbitrary commands or downstream execution, and natively resumable as the same unit; otherwise strict downgrade applies.
-- Direct reading, searching, editing, and testing by the current Main Agent remain allowed.
-- `review-report.md` is the only persisted review artifact; do not append separate Review Notes to `context-snapshot.md`.
-
-Missing required evidence, an unsafe repair, a failed verification, or failure to converge by round ten produces `NEEDS_FIXES`.
+Missing required evidence, an unsafe repair, a failed verification, or failure to converge by the contract's review limit produces `NEEDS_FIXES`.
 
 **Gate Check — Final Review Result**:
 

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

@@ -2,21 +2,13 @@
 
 Review the completed implementation before the full Feature test gate runs. A review result is a required predecessor; never accept or route around testing as if a review had passed.
 
-Use `artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` so `/prizmkit-code-review` can read the spec, plan, context snapshot, implementation log, and current diff. The skill runs a mandatory Main-Agent review loop of up to ten completed rounds and writes only `review-report.md`. After Main-Agent convergence, the skill may create one skill-owned independent Reviewer only when its strict structural capability gate passes; otherwise it records downgrade and creates none.
+Use `artifact_dir=.prizmkit/specs/{{FEATURE_SLUG}}/` so `/prizmkit-code-review` can read the spec, plan, context snapshot, implementation log, and current diff. Follow the skill's current review contract for review execution, repairs, verification, evidence, and handoff. Do not invoke another review skill or add a second review path outside that contract.
 
 - `REVIEW_PASS` → update checkpoint metadata with `stage=code-review`, `status=REVIEW_PASS`, `stage_result=REVIEW_PASS`, `next_stage=prizmkit-test`, and `resume_from=prizmkit-test`; then run the full test gate.
 - `REVIEW_NEEDS_FIXES` → preserve findings, increment the shared outer `repair_round`, route `prizmkit-implement` → `prizmkit-code-review` → `prizmkit-test`, and do not accept any test result from before the repaired review.
 - If the outer repair round is already three, record `WORKFLOW_BLOCKED`; do not start another repair route.
 
-Missing required evidence, unsafe repair, failed verification, or failure to converge by round ten produces `REVIEW_NEEDS_FIXES`.
-
-The Code Review execution boundary is closed:
-
-- Do not delegate the mandatory Main-Agent review or invoke another review skill or workflow.
-- Do not add review execution outside the single optional Reviewer owned and gated by `prizmkit-code-review`.
-- The optional Reviewer must be structurally read-only, unable to execute arbitrary commands or downstream execution, and natively resumable as the same unit; otherwise strict downgrade applies.
-- Direct reading, searching, editing, and testing by the current Main Agent remain allowed.
-- `review-report.md` is the only persisted review artifact; do not append separate Review Notes to `context-snapshot.md`.
+Missing required evidence, unsafe repair, failed verification, or failure to converge under the review contract produces `REVIEW_NEEDS_FIXES`.
 
 **Gate Check — Final Review Result**:
 

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

@@ -71,7 +71,7 @@ Before proceeding past CP-1, verify:
 
 ### Plan Self-Check — Orchestrator Direct
 
-Do not delegate planning analysis. The current Main Agent performs a bounded self-check before implementation:
+The active planning skill and runtime determine whether any additional planning execution is appropriate. Do not treat this bounded self-check as a substitute for the planning skill's own review contract.
 
 1. Re-read only `context-snapshot.md`, `spec.md`, and `plan.md`.
 2. Cross-check spec.md and plan.md Tasks against the feature description and Verification Gates in the Task Contract.

package/bundled/dev-pipeline/templates/sections/refactor-mission.md

@@ -4,6 +4,6 @@ You are the **refactor session orchestrator**. Execute Refactor {{REFACTOR_ID}}:
 
 **CRITICAL**: Preserve observable behavior. Do not change product behavior unless the refactor contract explicitly requires it.
 
-**Execution model**: You implement directly in the current checkout. Do not spawn a top-level Dev subagent. The current Main Agent runs `/prizmkit-code-review` directly and owns its complete review, repair, verification, and reporting loop without direct or indirect review delegation.
+**Execution model**: Follow the active skills and runtime contracts for planning, implementation, review, recovery, and commit. Those contracts determine the execution topology, permissions, and delegation boundaries.
 
 **NON-INTERACTIVE MODE**: There is NO human on the other end. Resolve uncertainty from existing code and conservative project patterns.

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

@@ -1,11 +1,11 @@
-### Phase 2: Implement — Orchestrator Direct
+### Phase 2: Implement — Plan Execution
 
 **Goal**: execute the refactor plan while preserving behavior.
 
 Run `/prizmkit-implement` directly with `artifact_dir=.prizmkit/refactor/{{REFACTOR_ID}}/`.
 
 Rules:
-- You are the implementer; do not spawn a top-level Dev subagent.
+- Follow the implementation skill's current execution and delegation contract.
 - Establish a baseline test result before structural changes when feasible.
 - Execute plan tasks in order and mark each `[x]` immediately.
 - After each behavior-sensitive task, run the smallest relevant behavior-preservation check.

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

@@ -1,16 +1,8 @@
 ### Phase 3: Review — Code Review & Behavior Verification
 
-Run `/prizmkit-code-review` directly with `artifact_dir=.prizmkit/refactor/{{REFACTOR_ID}}/`. The skill runs a mandatory Main-Agent review loop for up to ten completed rounds: the Main Agent reviews the complete current change, rejects unsupported candidate findings, directly repairs accepted findings, verifies repairs, and repeats until `accepted = 0` with no unresolved findings. After convergence, one skill-owned independent Reviewer is optional only when the strict structural capability gate passes; otherwise no Reviewer is created.
+Run `/prizmkit-code-review` directly with `artifact_dir=.prizmkit/refactor/{{REFACTOR_ID}}/` and follow the skill's complete review contract for review execution, repairs, verification, evidence, and handoff. Do not invoke another review skill or add a second review path outside that contract.
 
-The Code Review execution boundary is closed for this phase:
-
-- Do not delegate the mandatory Main-Agent review or invoke another review skill or workflow.
-- Do not add review execution outside the single optional Reviewer owned and gated by `prizmkit-code-review`.
-- The optional Reviewer must be structurally read-only, unable to execute arbitrary commands or downstream execution, and natively resumable as the same unit; otherwise strict downgrade applies.
-- Direct reading, searching, editing, and testing by the current Main Agent remain allowed.
-- `review-report.md` is the only persisted review artifact; do not append separate Review Notes to another artifact.
-
-Missing required evidence, an unsafe repair, a failed verification, or failure to converge by round ten produces `NEEDS_FIXES`.
+Missing required evidence, an unsafe repair, a failed verification, or failure to converge under the review contract produces `NEEDS_FIXES`.
 
 **Gate Check — Final Review Result**:
 

package/bundled/dev-pipeline/templates/sections/refactor-reminders.md

@@ -3,7 +3,8 @@
 - Use L1 skills: `/prizmkit-plan`, `/prizmkit-implement`, `/prizmkit-code-review`, `/prizmkit-retrospective`, `/prizmkit-committer`.
 - All refactor skills use `artifact_dir=.prizmkit/refactor/{{REFACTOR_ID}}/`.
 - Behavior preservation is the first priority.
-- Do not spawn a top-level Dev implementation subagent; Code Review keeps mandatory Main-Agent ownership and may use only its strict-gated skill-owned Reviewer after convergence.
+- Follow the active skills and runtime contracts; they determine execution topology, review ownership, and delegation boundaries
+- On execution-unit timeout, preserve the active checkout and artifacts and follow the owning contract's recovery or downgrade path
 - Main orchestrator runs `/prizmkit-code-review` directly.
 - Commit with `refactor(<scope>):`, not `feat:` or `fix:`.
 - Checkpoint state is the recovery truth; skipped steps must be marked `skipped`.

package/bundled/dev-pipeline/templates/sections/subagent-timeout-recovery.md

@@ -1,8 +1,9 @@
-## Normal Work Delegation Timeout Recovery
+## Execution-Unit Timeout Recovery
 
-This recovery path applies only to normal work delegated outside the optional skill-owned review contract. Code Review never applies this retry procedure to its independent Reviewer; Reviewer creation or native-resume failure uses the skill's strict downgrade instead.
+This recovery path applies to any delegated execution unit created by the active skill or runtime. The unit's own skill contract determines whether it may be resumed, replaced, or handled by the Main Agent; do not infer those rules from this prompt.
 
-If a normal work execution unit times out:
+If an execution unit times out:
 1. `{{RUNTIME_HELPER_CMD}} artifact list .prizmkit/specs/{{FEATURE_SLUG}}` — check what exists
 2. If `context-snapshot.md` exists: open a recovery prompt with `"Read .prizmkit/specs/{{FEATURE_SLUG}}/context-snapshot.md for project context and the Implementation Log. Run git diff HEAD to see actual code changes already made. Do NOT re-read individual source files unless the File Manifest directs you to."` + only remaining steps + `model: "lite"`
-3. Max 2 retries per phase. After 2 failures, the Main Agent completes the work directly and appends a Recovery Note to context-snapshot.md.
+3. Apply the active skill's retry and continuation contract. Do not create a replacement execution unit when that contract requires native continuation or strict downgrade.
+4. After the allowed recovery attempts are exhausted, the Main Agent continues directly when safe or records `failure-log.md` with the timeout and last observable state.

package/bundled/dev-pipeline/tests/test_generate_bootstrap_prompt.py

@@ -51,10 +51,8 @@ FORBIDDEN_THREE_STRIKE = "3-" + "strike"
 FORBIDDEN_TWENTY_STEP = "20-" + "step window"
 
 REVIEW_PROMPT_REQUIRED = (
-    "mandatory Main-Agent review",
-    "strict structural capability gate",
-    "skill-owned independent Reviewer",
-    "`review-report.md` is the only persisted review artifact",
+    "review-report.md",
+    "Final Result",
 )
 
 REVIEW_PROMPT_FORBIDDEN = (
@@ -246,7 +244,7 @@ def write_fresh_structured_evidence(project_root):
         "EVIDENCE_VALIDATE",
     ]
     for name in stage_names:
-        stages.append({"name": name, "status": "complete", "input_sha256": zero_hash, "outputs": []})
+        stages.append({"name": name, "status": "complete", "outputs": []})
     manifest = {
         "protocol_version": "1.0",
         "evidence_id": evidence_id,
@@ -1235,10 +1233,8 @@ class TestDirectOrchestratorImplementationPrompts:
             assert FORBIDDEN_TEST_CMD_PLACEHOLDER not in rendered
             assert FORBIDDEN_THREE_STRIKE not in rendered
             assert FORBIDDEN_TWENTY_STEP not in rendered
-            assert "mandatory Main-Agent review" in rendered
-            assert "up to ten completed rounds" in rendered
-            assert "strict structural capability gate" in rendered
-            assert "skill-owned independent Reviewer" in rendered
+            assert "review-report.md" in rendered
+            assert "Final Result" in rendered
             assert "low=0" not in rendered
             assert "Reviewer 3" not in rendered
 
@@ -1273,7 +1269,7 @@ class TestDirectOrchestratorImplementationPrompts:
         rendered = Path("dev-pipeline/templates/bootstrap-tier3.md").read_text(encoding="utf-8")
 
         assert "/prizmkit-implement" in rendered
-        assert "Implement — Orchestrator Direct" in rendered
+        assert "Implement — Plan Execution" in rendered
         assert "Spawn Dev subagent" not in rendered
         assert "Spawn Dev agent" not in rendered
         assert "Spawn Reviewer agent" not in rendered
@@ -1663,7 +1659,7 @@ class TestFeaturePromptUnifiedFullGuidance:
         prompt = _render_feature_prompt(tmp_path / "fallback-tier3", mode="lite", complexity="low")
 
         assert requested.index("bootstrap-tier3.md") < requested.index("bootstrap-prompt.md") if "bootstrap-prompt.md" in requested else True
-        assert "# Dev-Pipeline Session Bootstrap — Tier 3" in prompt
+        assert "# Dev-Pipeline Session Bootstrap — Full Guidance" in prompt
         assert "bootstrap-tier1.md" not in requested
         assert "bootstrap-tier2.md" not in requested
 
@@ -1931,9 +1927,8 @@ class TestHeadlessPromptCleanupF033:
         refactor_prompt = _render_refactor_prompt(tmp_path / "refactor-mixed", platform="claude")
 
         assert "{{" + "REVIEWER_" + "SUBAGENT_PATH" + "}}" not in replacements
-        assert "mandatory Main-Agent review" in refactor_prompt
-        assert "strict structural capability gate" in refactor_prompt
-        assert "skill-owned independent Reviewer" in refactor_prompt
+        assert "review-report.md" in refactor_prompt
+        assert "Final Result" in refactor_prompt
         assert "Reviewer 3" not in refactor_prompt
 
     def test_bugfix_refactor_explicit_codex_override(self, tmp_path, monkeypatch):
@@ -1955,9 +1950,8 @@ class TestHeadlessPromptCleanupF033:
         refactor_prompt = _render_refactor_prompt(tmp_path / "refactor-codex", platform="claude", explicit_platform="codex")
 
         assert "{{" + "REVIEWER_" + "SUBAGENT_PATH" + "}}" not in replacements
-        assert "mandatory Main-Agent review" in refactor_prompt
-        assert "strict structural capability gate" in refactor_prompt
-        assert "skill-owned independent Reviewer" in refactor_prompt
+        assert "review-report.md" in refactor_prompt
+        assert "Final Result" in refactor_prompt
         assert "Reviewer 3" not in refactor_prompt
 
 

package/bundled/dev-pipeline/tests/test_generate_bugfix_prompt.py

@@ -251,7 +251,9 @@ class TestHeadlessRecoveryReframing:
     def test_bugfix_template_does_not_spawn_monitor_or_require_compact(self):
         content = Path("dev-pipeline/templates/bugfix-bootstrap-prompt.md").read_text(encoding="utf-8")
 
-        assert "Do not spawn persistent background subagents for headless recovery" in content
+        assert "On execution-unit timeout, preserve the active checkout and artifacts" in content
+        assert "follow the owning contract's recovery or downgrade path" in content
+        assert "Do not spawn persistent background subagents for headless recovery" not in content
         assert REMOVED_MAX_LOG_ENV not in content
         assert REMOVED_MAX_LOG_HUMAN not in content
         assert ("209" + "7152") not in content

package/bundled/dev-pipeline/tests/test_python_runner_parity.py

@@ -1007,7 +1007,6 @@ def test_prompt_generation_result_preserves_per_item_model(monkeypatch, tmp_path
                     "checkpoint_path": str(paths.project_root / ".prizmkit" / "bugfix" / "B-001" / "workflow-checkpoint.json"),
                     "model": "per-item-model",
                     "pipeline_mode": "standard",
-                    "agent_count": 3,
                 }
             ),
             stderr="",
@@ -1056,7 +1055,6 @@ def test_prompt_generation_passes_explicit_execution_root_to_generator(monkeypat
                     "checkpoint_path": str(execution_root / ".prizmkit" / "specs" / "001-low" / "workflow-checkpoint.json"),
                     "model": "",
                     "pipeline_mode": "full",
-                    "agent_count": 3,
                 }
             ),
             stderr="",
@@ -1121,7 +1119,6 @@ def _install_runner_session_fakes(monkeypatch, runners, *, statuses=("success",)
             output_path=session_paths.prompt_path,
             model="per-item-model",
             pipeline_mode="full",
-            agent_count=3,
             checkpoint_path=execution_root / ".prizmkit" / "specs" / "001-low" / "workflow-checkpoint.json",
             artifact_path=execution_root / ".prizmkit" / "specs" / "001-low",
             raw={},
@@ -1445,7 +1442,6 @@ def test_explicit_worktree_materializes_ignored_support_assets_before_prompt_and
             output_path=session_paths.prompt_path,
             model="",
             pipeline_mode="full",
-            agent_count=3,
             checkpoint_path=execution_root / ".prizmkit" / "specs" / "001-support" / "workflow-checkpoint.json",
             artifact_path=execution_root / ".prizmkit" / "specs" / "001-support",
             raw={},
@@ -1910,8 +1906,7 @@ def test_foreground_runner_emits_structured_feature_progress(monkeypatch, tmp_pa
     assert "Feature: F-001 — Readable output" in captured.err
     assert "Code retry: 1 / 3" in captured.err
     assert "Preflight: no user-visible changes to preserve before feature task F-001" in captured.err
-    assert "Pipeline mode: full (Tier 3 — Main-Agent Full Guardrails)" in captured.err
-    assert "Agents: 3" in captured.err
+    assert "Pipeline mode: full (Full Guidance)" in captured.err
     assert "Spawning AI CLI session: F-001-session-1" in captured.err
     assert "Session result: success" in captured.err
 

package/bundled/dev-pipeline/tests/test_unified_cli.py

@@ -812,7 +812,7 @@ def test_session_preamble_writes_redacted_start_command_for_all_launch_profiles(
             assert str(prompt) not in "\n".join(lines[:6])
 
 
-def test_headless_review_guidance_uses_main_agent_review_loop():
+def test_headless_review_guidance_defers_to_skill_review_contract():
     headless_guidance_files = [
         REPO_ROOT / "dev-pipeline" / "templates" / "sections" / "phase-review-full.md",
         REPO_ROOT / "dev-pipeline" / "templates" / "bootstrap-tier3.md",
@@ -828,11 +828,9 @@ def test_headless_review_guidance_uses_main_agent_review_loop():
 
     for path in headless_guidance_files:
         text = path.read_text(encoding="utf-8")
-        assert "mandatory Main-Agent review" in text, path
-        assert "up to ten completed rounds" in text, path
+        assert "current review contract" in text, path
+        assert "second review path outside that contract" in text, path
         assert "text final-verdict" in text, path
-        assert "strict structural capability gate" in text, path
-        assert "skill-owned independent Reviewer" in text, path
         assert "low=0" not in text and "Reviewer 3" not in text, path
         assert "forced-worktree changed-content fallback" not in text, path
         assert "isolation:" not in text, path

package/bundled/skills/_metadata.json

@@ -1,5 +1,5 @@
 {
-  "version": "1.1.136",
+  "version": "1.1.138",
   "skills": {
     "prizmkit": {
       "description": "Framework introduction and navigation for the formal single-requirement lifecycle, project initialization, Prizm docs, and independent deployment.",
@@ -72,7 +72,7 @@
       "hasScripts": false
     },
     "prizmkit-test": {
-      "description": "Mandatory post-review auditable test evidence with TEST_PASS/TEST_FAIL/TEST_BLOCKED verdicts, repair-scope routing, differential proof, replay, and deterministic validation.",
+      "description": "Mandatory post-review auditable test evidence with TEST_PASS/TEST_FAIL/TEST_BLOCKED verdicts, repair-scope routing, current-workspace execution evidence, replay, and deterministic validation.",
       "tier": "1",
       "category": "prizmkit-skill",
       "hasAssets": true,

package/bundled/skills/prizmkit-test/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: "prizmkit-test"
-description: "Auditable test-evidence protocol for AI-generated behavior changes. Classifies changes, discovers affected modules and regression rings, models observable behavior and risk, builds project-native tests and contract-driven mocks, proves tests differentially, preserves content-addressed raw evidence, and deterministically returns TEST_PASS, TEST_FAIL, or TEST_BLOCKED. Use after code/config/schema/dependency changes or when users ask to test, verify, add complete module tests, inspect boundaries, or produce auditable evidence. Supports full-project, module, feature, and scope=this-change. (project)"
+description: "Auditable test-evidence protocol for AI-generated behavior changes. Classifies changes, discovers affected modules and regression rings, models observable behavior and risk, builds project-native tests and contract-driven mocks, validates current-workspace test evidence, preserves content-addressed raw evidence, and deterministically returns TEST_PASS, TEST_FAIL, or TEST_BLOCKED. Use after code/config/schema/dependency changes or when users ask to test, verify, add complete module tests, inspect boundaries, or produce auditable evidence. Supports full-project, module, feature, and scope=this-change. (project)"
 ---
 
 # PrizmKit Test
@@ -11,16 +11,16 @@ This skill orchestrates each target project's native runners and conventions. It
 
 ## Model Choice vs Fixed Evidence Mechanics
 
-Choose project-semantic parameters from the target project: framework, commands, cwd, timeout, retry/attempt policy, concurrency, Mock tools, test layers, module boundary, inventory patterns, mutation technology, and environment classification. Do not hard-code ecosystem-specific defaults when the project can decide them.
+Choose project-semantic parameters from the target project: framework, commands, cwd, timeout, retry/attempt policy, concurrency, Mock tools, test layers, module boundary, inventory patterns, and environment classification. Do not hard-code ecosystem-specific defaults when the project can decide them.
 
-Fix only the protocol mechanisms needed for safe, replayable evidence: project/evidence/request locators, schema-shaped requests, path confinement, real process execution, complete raw capture, runner-generated chained receipts, hash binding, append-only history, differential isolation/cleanup, and resume invalidation. Read `${SKILL_DIR}/references/trusted-evidence-execution.md` before inventory, execution, differential proof, resume, or replay; it defines the trust boundary and builder commands. Read `${SKILL_DIR}/references/evidence-request-protocol.md` when authoring any request or authoritative JSON record; it defines request ownership, structured N/A, and the schema map.
+Fix only the protocol mechanisms needed for safe, replayable evidence: project/evidence/request locators, schema-shaped requests, path confinement, real process execution, complete raw capture, runner-generated receipts, append-only history, and resume invalidation. Read `${SKILL_DIR}/references/trusted-evidence-execution.md` before inventory, execution, resume, or replay; it defines the trust boundary and builder commands. Read `${SKILL_DIR}/references/evidence-request-protocol.md` when authoring any request or authoritative JSON record; it defines request ownership, structured N/A, and the schema map.
 
 ## Canonical Evidence Builder Lifecycle
 
 The Main Agent supplies semantic requests only; it does not author authoritative package records. Initialize and drive one evidence directory through the builder in this order:
 
 ```text
-init → capture-change → inventory → prepare-tests → execute/differential → finalize → render-report → validate --attest → resume
+init → capture-change → inventory → prepare-tests → execute → finalize → render-report → validate --attest → resume
 ```
 
 Use `${SKILL_DIR}/scripts/build_test_evidence.py` for every lifecycle command:
@@ -29,13 +29,12 @@ Use `${SKILL_DIR}/scripts/build_test_evidence.py` for every lifecycle command:
 - `capture-change`: captures the complete canonical change, including untracked, deleted, renamed, and copied files; never substitute plain `git diff`. `inventory` then requires its changed-file set to match the capture exactly and rejects stale captures or unaccounted live paths.
 - `inventory`: hashes current source/tests/contracts/lockfiles, rejects category overlap, excludes runtime-only paths, and validates directory roots plus the complete capture-derived changed-file set.
 - `prepare-tests`: discovers repository-native test files from the canonical capture and inventory, generates stable builder-owned test IDs, derives existing/added/modified status, and snapshots only added/modified live test sources.
-- `execute`: runs native project commands and owns `executions.json`, `receipts/`, and `raw/`. For `auto_bind=true`, the builder resolves recognized package-wide commands or explicit file/directory/glob arguments to the prepared tests they cover; ambiguous commands are blocked and must use `test_paths`.
-- `differential`: records executable proof or a mechanically validated `NOT_APPLICABLE` decision only for genuinely new behavior or textual-contract behavior.
-- `finalize`: owns manifest, stage hashes, verdict, generated-test snapshots, and evidence identity.
+- `execute`: runs native project commands and owns `executions.json`, `receipts/`, and `raw/`. Each request uses exactly one binding mode: `test_ids`, `test_paths`, or `auto_bind=true`. The builder infers recognized package-wide or explicit file/directory/glob command scope, blocks opaque commands or mismatched declarations, and persists verified canonical `test_ids` only.
+- `finalize`: rechecks canonical capture freshness, blocks every selected unreliable execution, and owns manifest, stage hashes, verdict, generated-test snapshots, and evidence identity.
 - `render-report`: derives `test-report.md`; it is never authoritative.
 - `resume`: preserves prior receipts and refuses to rebind an identity-changing package; create a new package after scope/source identity changes.
 
-`test-plan.json` is the only source of builder-generated test IDs. Execution requests may provide native `test_paths`, `auto_bind=true`, or legacy IDs; the builder resolves paths to IDs before execution and rejects unknown/unbound tests or layer mappings. Assess all five layers explicitly, but set `required=true` only when risk requires it; every omitted layer needs structured N/A evidence. The same binding applies test ↔ behavior ↔ execution ↔ differential proof.
+`test-plan.json` is the only source of builder-generated test IDs. An execution request provides exactly one of native `test_paths`, `auto_bind=true`, or canonical `test_ids`; the actual command must expose the same deterministic scope, including all prepared tests for recognized package-wide commands. The builder blocks opaque or mismatched command scope and rewrites verified requests to canonical `test_ids` only before execution. Assess all five layers explicitly, but set `required=true` only when risk requires it; every omitted layer needs structured N/A evidence. The same binding applies test ↔ behavior ↔ execution; current-workspace receipts are the execution evidence.
 
 The builder/validator, not the Main Agent, produces or updates `manifest.json`, `executions.json`, `receipts/*`, `validation.json`, `verdict.json`, `test-report.md`, and `source-change.patch`. `validation.json` is produced only by `validate_test_evidence.py --attest`; a package is not a `TEST_PASS` until the second strict validation succeeds.
 
@@ -174,18 +173,16 @@ If a valid test reveals a business/contract defect, do not fix production behavi
 
 ### 7. EXECUTE_PROVE
 
-Run required layers in plan order through `${SKILL_DIR}/scripts/build_test_evidence.py execute`. Use `test_paths` or `auto_bind=true` for new requests; the builder resolves and persists canonical test IDs before the native command runs. The model selects project-native argv, cwd, timeout, attempts, concurrency, layer, environment classification, and tool probes in schema-shaped requests. The runner performs the actual process execution and appends complete environment/tool versions, raw stdout/stderr hashes, request hash, runner hash, unique receipt, and receipt-chain binding to `executions.json`. Do not accept caller-authored execution JSON and do not retry until green.
+Run required layers in plan order through `${SKILL_DIR}/scripts/build_test_evidence.py execute`. Use exactly one of `test_paths`, `auto_bind=true`, or canonical `test_ids`; each must exactly match a deterministically inferable native command scope. The builder blocks opaque or mismatched commands and persists verified canonical test IDs before the native command runs. The model selects project-native argv, cwd, timeout, attempts, concurrency, layer, environment classification, and tool probes in schema-shaped requests. The runner performs the actual process execution and appends complete environment/tool versions, raw stdout/stderr hashes, request hash, runner hash, and unique receipt to `executions.json`. Do not accept caller-authored execution JSON and do not retry until green.
 
-Differentially prove each added/changed necessary behavior test:
+Validate the current implementation directly in the project root:
 
-- Prefer baseline failure plus current success in an isolated/uncontaminated environment.
-- If baseline is inapplicable, use a minimal controlled mutation tied to the same risk.
-- Bind proof to runner receipts, baseline commit, mutation apply/restore hashes, and complete cleanup. The builder isolates tests to temporary directories so live-workspace drift cannot interfere with differential execution.
-- Keep source identity separate from evidence integrity: the change capture (patch + changed_files) detects behavior-bearing project drift, while `manifest.json` hashes every evidence file and detects report/output tampering.
-- Completely remove mutations and verify the isolated current copy is unchanged. Never rebind completed proof to the mutable live project tree.
-- Classify each proof as `PROVEN`, structured `NOT_APPLICABLE`, or `UNPROVEN`.
+- Run every required layer through the project-native runner; do not create historical checkouts or mutate a second copy of the repository.
+- Bind each applicable behavior risk to planned tests and selected, reliable execution receipts. A passing current-state test is valid evidence even when the historical implementation would also have passed it.
+- Preserve complete runner receipts, raw outputs, request bindings, and replay links. The builder owns execution provenance; the Main Agent does not author receipt records.
+- Required test failures produce `TEST_FAIL`; any selected unreliable execution, missing mapping, unavailable execution, unresolved risk, or budget truncation produces `TEST_BLOCKED`. `TEST_PASS` requires every selected execution to be reliable with exit code zero and every planned test to have a selected reliable zero-exit receipt.
 
-Necessary `UNPROVEN` behavior, required flakiness, unavailable infrastructure after bounded recovery, failed cleanup, unreliable execution, or budget truncation yields `TEST_BLOCKED`.
+TDD history, baseline-failure proof, and mutation sensitivity are outside this skill. This skill validates whether the current implementation satisfies the modeled behavior and risk evidence.
 
 ### 8. EVIDENCE_PACKAGE
 
@@ -197,7 +194,6 @@ Create `.prizmkit/test/evidence/<evidence-id>/` from the package template with:
 - `behavior-risk-matrix.json`
 - `test-plan.json`
 - `infrastructure-changes.json`
-- `differential-proof.json`
 - `executions.json`
 - `verdict.json`
 - `source-change.patch`
@@ -220,7 +216,7 @@ python3 ${SKILL_DIR}/scripts/validate_test_evidence.py \
   --project-root <target-project-root>
 ```
 
-If this pre-check finds errors, fix them in the package before the second pass. The validator checks every authoritative record against its shipped schema, live target inventories, patch/diff binding, stage dependencies, changed-file/module-root/exclusion/Regression Ring cross-links, generated-test snapshots, planned-test execution links, runner-generated receipts/request/raw-output hashes, successful behavior-risk mappings, proof-linked isolated failures/current successes, production/unknown external-target safety, cleanup, and package integrity.
+If this pre-check finds errors, fix them in the package before the second pass. The validator checks every authoritative record against its shipped schema, live target inventories, patch/diff binding, changed-file/module-root/exclusion/Regression Ring cross-links, generated-test snapshots, planned-test execution links, runner-generated receipts/request/raw-output hashes, successful behavior-risk mappings, production/unknown external-target safety, and package integrity.
 
 **Second pass (attestation, with `--attest`):**
 
@@ -231,7 +227,7 @@ python3 ${SKILL_DIR}/scripts/validate_test_evidence.py \
   --attest
 ```
 
-`--attest` writes an integrity/protocol attestation into `validation.json`, re-derives `test-report.md`, re-hashes final records in `manifest.json`, then runs a **second** strict validation against the refreshed package. A package is not a `TEST_PASS` until this second strict validation succeeds. Use builder replay when feasible to rerun recorded requests and create new linked receipts. Store validation output. Tampering, schema/identity/hash mismatch, source/test/lockfile drift, missing matrix mappings, wrong execution order, invalid differential proof, cleanup failure, unjustified structured N/A, unresolved risk, or a false real-environment claim fails validation.
+`--attest` writes an integrity/protocol attestation into `validation.json`, re-derives `test-report.md`, re-hashes final records in `manifest.json`, then runs a **second** strict validation against the refreshed package. A package is not a `TEST_PASS` until this second strict validation succeeds. Use builder replay when feasible to rerun recorded requests and create new linked receipts. Store validation output. Tampering, schema/identity/hash mismatch, source/test/lockfile drift, missing matrix mappings, wrong execution order, unjustified structured N/A, unresolved risk, or a false real-environment claim fails validation.
 
 ## Resume and Idempotency
 
@@ -249,9 +245,9 @@ For a large module or bounded budget, persist matrix/checkpoint progress and ret
 
 Only these testing-domain outcomes are valid:
 
-- `TEST_PASS`: every required module behavior/risk and Regression Ring check is proven, required executions pass reliably, cleanup succeeds, no verdict-capable unresolved edge remains, and deterministic validation passes.
+- `TEST_PASS`: every required module behavior/risk and Regression Ring check is covered by current-workspace tests, required executions pass reliably, no verdict-capable unresolved edge remains, and deterministic validation passes.
 - `TEST_FAIL`: a valid reliable test reproduces an implementation or resolved-contract failure. The authoritative verdict record must include `repair_scope` and enough evidence for the next stage to route safely.
-- `TEST_BLOCKED`: unknown scope, conflicting truth, unavailable/unreliable/flaky necessary execution, incomplete evidence, unproven tests, failed cleanup, budget truncation, or deterministic validation failure.
+- `TEST_BLOCKED`: unknown scope, conflicting truth, unavailable/unreliable/flaky necessary execution, incomplete evidence, unmapped tests, budget truncation, or deterministic validation failure.
 
 No conditional pass exists.