rungate 0.2.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (196) hide show
  1. package/.claude/agents/aditi.md +47 -0
  2. package/.claude/agents/marcus.md +50 -0
  3. package/.claude/agents/quinn.md +58 -0
  4. package/.claude/agents/rook.md +47 -0
  5. package/.claude/agents/serena.md +46 -0
  6. package/.claude/rungate.json +11 -0
  7. package/.fallowrc.json +28 -0
  8. package/.github/copilot-instructions.md +20 -0
  9. package/.rungate/conformity-findings.json +648 -0
  10. package/AGENTS.md +106 -0
  11. package/CODE-MAP.md +50 -0
  12. package/HARNESS.md +88 -0
  13. package/bun.lock +26 -0
  14. package/bunfig.toml +2 -0
  15. package/config/brief-policies.json +28 -0
  16. package/config/ceremony-profiles.json +212 -0
  17. package/config/consumer-cascade-map.json +24 -0
  18. package/config/doc-cascade-map.json +24 -0
  19. package/config/eval.json +84 -0
  20. package/config/profiles/account-plan.yaml +36 -0
  21. package/config/profiles/campaign.yaml +94 -0
  22. package/config/prove-evidence.schema.json +65 -0
  23. package/config/spec-policies.json +36 -0
  24. package/config/workflow-schema.json +281 -0
  25. package/gates/SCHEMA-GUIDE.md +121 -0
  26. package/gates/ac-quality.test.ts +106 -0
  27. package/gates/adversarial.test.ts +274 -0
  28. package/gates/brief-assembler.test.ts +429 -0
  29. package/gates/brief-assembler.ts +334 -0
  30. package/gates/bunfig.toml +2 -0
  31. package/gates/chain.test.ts +91 -0
  32. package/gates/e2e-smoke.test.ts +234 -0
  33. package/gates/error-classifier.test.ts +249 -0
  34. package/gates/error-classifier.ts +89 -0
  35. package/gates/orchestrator.test.ts +950 -0
  36. package/gates/orchestrator.ts +569 -0
  37. package/gates/preload.ts +77 -0
  38. package/gates/prompts/ac-adversary.md +20 -0
  39. package/gates/prompts/evidence-validator.md +22 -0
  40. package/gates/prompts/prove-reproducer.md +53 -0
  41. package/gates/prove.test.ts +164 -0
  42. package/gates/run-gate.ts +936 -0
  43. package/gates/schema-parity.test.ts +197 -0
  44. package/gates/schema.ts +362 -0
  45. package/gates/self-heal.test.ts +511 -0
  46. package/gates/self-heal.ts +121 -0
  47. package/gates/ship-orchestrator.test.ts +379 -0
  48. package/gates/ship-orchestrator.ts +363 -0
  49. package/gates/test-fixtures/broken-no-environments.json +60 -0
  50. package/gates/test-fixtures/valid-standard.json +54 -0
  51. package/gates/test-utils.ts +43 -0
  52. package/gates/witness.ts +220 -0
  53. package/gates/workflow.test.ts +1053 -0
  54. package/hooks/AgentBriefGuard.hook.ts +586 -0
  55. package/hooks/AgentVerdictCapture.hook.ts +134 -0
  56. package/hooks/AutoVerifyGate.hook.ts +76 -0
  57. package/hooks/GateEnforcement.hook.ts +273 -0
  58. package/hooks/IssueCloseGuard.hook.ts +116 -0
  59. package/hooks/MergeGuard.hook.ts +55 -0
  60. package/hooks/StaleTTLCleanup.hook.ts +133 -0
  61. package/hooks/VerifyPhaseLock.hook.ts +81 -0
  62. package/hooks/WorkflowStateGuard.hook.ts +37 -0
  63. package/hooks/lib/agentDetection.ts +70 -0
  64. package/hooks/lib/findWorkflow.ts +101 -0
  65. package/hooks/lib/parseStdin.ts +37 -0
  66. package/hooks/lib/paths.ts +111 -0
  67. package/hooks/lib/utils.ts +81 -0
  68. package/lib/conformity.ts +959 -0
  69. package/lib/paths.ts +23 -0
  70. package/lib/rungate-schema.ts +60 -0
  71. package/lib/signal-phrases.ts +7 -0
  72. package/package.json +20 -0
  73. package/prompts/ac-adversary.md +20 -0
  74. package/prompts/ac-format.md +34 -0
  75. package/prompts/aditi.md +29 -0
  76. package/prompts/blast-radius.md +29 -0
  77. package/prompts/coding-principles.md +34 -0
  78. package/prompts/container-rebuild.md +29 -0
  79. package/prompts/container-verify.md +28 -0
  80. package/prompts/discovery.md +27 -0
  81. package/prompts/environment.md +28 -0
  82. package/prompts/escalation-decision-tree.md +38 -0
  83. package/prompts/evidence-hierarchy.md +31 -0
  84. package/prompts/evidence-validator.md +22 -0
  85. package/prompts/marcus.md +29 -0
  86. package/prompts/prevention.md +26 -0
  87. package/prompts/prove-reproducer.md +43 -0
  88. package/prompts/quinn-decision-tree.md +31 -0
  89. package/prompts/quinn-ui-brief.md +106 -0
  90. package/prompts/quinn.md +29 -0
  91. package/prompts/rca.md +28 -0
  92. package/prompts/read-before-write.md +30 -0
  93. package/prompts/regression.md +26 -0
  94. package/prompts/rook.md +29 -0
  95. package/prompts/serena.md +28 -0
  96. package/prompts/testing-strategy.md +34 -0
  97. package/reference/migration-manifest.json +554 -0
  98. package/reference/prompts/agent-aditi.md +38 -0
  99. package/reference/prompts/agent-marcus.md +38 -0
  100. package/reference/prompts/agent-quinn.md +38 -0
  101. package/reference/prompts/agent-rook.md +37 -0
  102. package/reference/prompts/agent-serena.md +37 -0
  103. package/reference/prompts/council-skill.md +170 -0
  104. package/reference/prompts/goal-skill.md +235 -0
  105. package/reference/prompts/prove-skill.md +283 -0
  106. package/reference/prompts/quinn-standard.md +246 -0
  107. package/reference/prompts/ship-convergence.md +65 -0
  108. package/reference/prompts/ship-escalation.md +43 -0
  109. package/reference/prompts/ship-iteration.md +382 -0
  110. package/reference/prompts/ship-pending-patches.md +246 -0
  111. package/reference/prompts/ship-reference.md +250 -0
  112. package/reference/prompts/ship-skill.md +347 -0
  113. package/reference/prompts/ship-templates.md +215 -0
  114. package/reference/scripts/batch-status.sh +68 -0
  115. package/reference/scripts/chain-compat-check.sh +67 -0
  116. package/reference/scripts/context-audit.sh +314 -0
  117. package/reference/scripts/dead-export-check.sh +61 -0
  118. package/reference/scripts/generate-qc-tables.sh +92 -0
  119. package/reference/scripts/harness-init.sh +111 -0
  120. package/reference/scripts/pai-054-baseline-check.sh +40 -0
  121. package/reference/scripts/prove-smoke-test.sh +112 -0
  122. package/reference/scripts/setup-keychain-salt.sh +12 -0
  123. package/reference/scripts/skill-optimizer-friday.sh +88 -0
  124. package/reference/scripts/skill-qc-test.sh +351 -0
  125. package/reference/scripts/skill-synthesize-tonight.sh +66 -0
  126. package/reference/specs/HARNESS-EXTRACTION-SPEC.md +263 -0
  127. package/reference/specs/harness-v3-bash-deletion-plan.md +65 -0
  128. package/reference/specs/harness-v3-migration-plan.md +188 -0
  129. package/scripts/decision-reconcile.sh +74 -0
  130. package/scripts/decision-reconcile.ts +152 -0
  131. package/scripts/extract-constraints.ts +260 -0
  132. package/scripts/generate-code-map.ts +339 -0
  133. package/scripts/git-hooks/pre-commit +76 -0
  134. package/scripts/git-hooks/pre-push +271 -0
  135. package/scripts/harness-skill-check.sh +52 -0
  136. package/scripts/lib/gap-emit.sh +65 -0
  137. package/scripts/lib/jq-update.sh +30 -0
  138. package/scripts/lib/resolve-slug.sh +30 -0
  139. package/scripts/lib/verify-spec-compliance.sh +102 -0
  140. package/scripts/prove-backfill.sh +87 -0
  141. package/scripts/scaffold-project.ts +1488 -0
  142. package/scripts/scaffold-rungate-config.ts +91 -0
  143. package/scripts/self-containment-check.sh +109 -0
  144. package/scripts/skill-qc-validator.sh +402 -0
  145. package/scripts/slug-resolver.sh +14 -0
  146. package/scripts/sync-spec-tests.ts +236 -0
  147. package/specs/BOOTSTRAP-DATA-FLOW-SPEC.md +2493 -0
  148. package/specs/BOOTSTRAP-TEST-PLAN.md +571 -0
  149. package/specs/HARNESS-GATES.md +256 -0
  150. package/specs/HARNESS-SKILL-CHAIN.md +286 -0
  151. package/specs/HARNESS-SKILL-CONTRACT.md +93 -0
  152. package/specs/HARNESS-STANDARD.md +488 -0
  153. package/specs/SPEC-TEMPLATE.md +53 -0
  154. package/specs/bootstrap-data-flow.png +0 -0
  155. package/specs/harness-automation-matrix.md +314 -0
  156. package/specs/harness-automation-matrix.rules.json +64 -0
  157. package/test/anti.test.ts +80 -0
  158. package/test/contract-negative.test.ts +86 -0
  159. package/test/contract.test.ts +158 -0
  160. package/test/external-deps.test.ts +48 -0
  161. package/test/fixtures/golden-project/.env.example +3 -0
  162. package/test/fixtures/golden-project/Makefile +7 -0
  163. package/test/fixtures/golden-project/docs/adr/ADR-001-framework.md +23 -0
  164. package/test/fixtures/golden-project/docs/setup-guide.md +10 -0
  165. package/test/fixtures/golden-project/package.json +10 -0
  166. package/test/fixtures/golden-project/specs/api-spec.md +20 -0
  167. package/test/fixtures/golden-project/src/auth.ts +8 -0
  168. package/test/fixtures/golden-project/src/db.ts +11 -0
  169. package/test/fixtures/golden-project/src/index.ts +19 -0
  170. package/test/fixtures/golden-project/src/utils/format.ts +3 -0
  171. package/test/fixtures/golden-project/test/auth.test.ts +15 -0
  172. package/test/fixtures/golden-project/tsconfig.json +12 -0
  173. package/test/meta-sc-coverage.test.ts +159 -0
  174. package/test/phase-0.test.ts +846 -0
  175. package/test/phase-1.test.ts +617 -0
  176. package/test/phase-2.test.ts +620 -0
  177. package/test/phase-3.test.ts +99 -0
  178. package/test/scaffold-conformity.test.ts +8 -0
  179. package/test/schema-canary.test.ts +51 -0
  180. package/test/spec-compliance-auto.test.ts +64 -0
  181. package/test/spec-compliance.test.ts +295 -0
  182. package/test/spec-discovery.test.ts +68 -0
  183. package/test/spec-drift.test.ts +5 -0
  184. package/test/structure.test.ts +165 -0
  185. package/test/unit/agent-brief-guard-ship.test.ts +293 -0
  186. package/test/unit/extract-constraints.test.ts +145 -0
  187. package/test/unit/issue-close-guard.test.ts +204 -0
  188. package/test/unit/scaffold-project.test.ts +417 -0
  189. package/test/unit/skill-cooccurrence.test.ts +177 -0
  190. package/test/unit/skill-sequence-logger.test.ts +244 -0
  191. package/test/unit/workflow-state-guard.test.ts +95 -0
  192. package/workflows/batch-ship.js +295 -0
  193. package/workflows/council.js +481 -0
  194. package/workflows/prove.js +560 -0
  195. package/workflows/ship.js +789 -0
  196. package/workflows/verify.js +236 -0
@@ -0,0 +1,215 @@
1
+ ---
2
+ doc-type: reference
3
+ status: active
4
+ owner: jason
5
+ updated: 2026-07-28
6
+ ---
7
+
8
+ # Ship Templates
9
+
10
+ Structured templates for artifacts produced during ship. Gates validate these structures exist on the issue.
11
+
12
+ ---
13
+
14
+ ## AC Template — CODE (8 required fields + 1 optional)
15
+
16
+ ```markdown
17
+ ## AC-N
18
+ - **Type:** REPLACE:CODE_OR_OUTCOME
19
+ - **Statement:** REPLACE:specific_verifiable_statement
20
+ - **Call Site:** REPLACE:file_and_line_where_this_is_called_from_and_trigger_event
21
+ - **Metric:** REPLACE:measurable_count_or_boolean
22
+ - **Threshold:** REPLACE:number_or_explicit_boolean
23
+ - **Baseline:** REPLACE:current_measured_value
24
+ - **Evidence Method:** REPLACE:grep_curl_screenshot_or_test
25
+ - **Pass Criteria:** REPLACE:exact_condition_that_proves_pass
26
+ - **Source:** REPLACE:architecture_adr_or_issue_reference
27
+ - **Data Flow:** REPLACE:state_file_path | verification_command (OPTIONAL — for wiring/integration ACs only)
28
+ ```
29
+
30
+ > **Data Flow rule:** For ACs that wire cross-system pipelines (hook A calls function B, pipeline feeds downstream state), include the Data Flow field. Format: `state_file_path | shell_command_that_returns_0_on_success`. verify-gate runs the command and checks the state file was modified. Omit for pure logic, test, or config ACs. Example: `~/.agentgrit/state/rule-stats.json | bun -e 'require("./src/capture/rating").parseRating("/rate M:4 S:4 Q:4")' `
31
+
32
+ > **Call Site rule:** Every new function must specify WHERE it is called from (file:line) and WHEN it fires (trigger event). A function with tests but no call site is dead code. Evidence must show the call site exists via grep. If the AC is for a test-only or config change with no new callable function, use `Call Site: N/A — no new function`.
33
+
34
+ ## AC Template — OUTCOME (10 required fields)
35
+
36
+ ```markdown
37
+ ## AC-N
38
+ - **Type:** REPLACE:CODE_OR_OUTCOME
39
+ - **Statement:** REPLACE:what_user_sees_or_gets
40
+ - **Metric:** REPLACE:count_percentage_time_or_field_presence
41
+ - **Threshold:** REPLACE:specific_number_or_condition
42
+ - **Baseline:** REPLACE:current_state_or_NA_for_new
43
+ - **Evidence Method:** REPLACE:curl_command_screenshot_or_playwright
44
+ - **Pass Criteria:** REPLACE:what_evidence_output_must_show
45
+ - **Source:** REPLACE:principles_goal_or_issue_reference
46
+ - **User:** REPLACE:who_uses_this
47
+ - **Conversation:** REPLACE:what_this_enables
48
+ - **Action:** REPLACE:specific_verb_and_context
49
+ ```
50
+
51
+ ---
52
+
53
+ ## ATTEMPT Comment Template
54
+
55
+ Posted to issue after every failed iteration. Required fields enforced by iteration-gate.
56
+
57
+ ```markdown
58
+ ## ATTEMPT N
59
+ - **Approach:** REPLACE:specific_files_and_changes
60
+ - **Files Changed:** REPLACE:list_of_files_with_changes
61
+ - **Result:** FAIL
62
+ - **Evidence:** REPLACE:curl_output_screenshot_or_error
63
+ - **Why It Failed:** REPLACE:root_cause_or_unknown
64
+ ```
65
+
66
+ ---
67
+
68
+ ## Sizing Declaration Template
69
+
70
+ Posted to issue during SCOPE. Compared to actuals at CLOSE.
71
+
72
+ ```markdown
73
+ ## Sizing Declaration
74
+ - **Predicted:** REPLACE:XS_S_M_or_L
75
+ - **Files:** REPLACE:count_and_names
76
+ - **Reasoning:** REPLACE:why_this_size
77
+ - **Expected time:** REPLACE:hours
78
+ - **Expected boundaries:** REPLACE:directories_or_none
79
+ ```
80
+
81
+ ---
82
+
83
+ ## Sizing Outcome Template
84
+
85
+ Posted to issue at CLOSE. Compared to declaration for accuracy tracking.
86
+
87
+ ```markdown
88
+ ## Sizing Outcome
89
+ - **Declared:** REPLACE:size_from_declaration
90
+ - **Actual:** REPLACE:actual_files_changed_count
91
+ - **Actual directories:** REPLACE:unique_top_level_dir_count
92
+ - **Actual time:** REPLACE:hours_from_checkpoint
93
+ - **Mismatch:** REPLACE:YES_or_NO
94
+ - **Docs read:** REPLACE:count_and_names
95
+ - **Lines loaded:** REPLACE:total_lines_across_all_docs
96
+ ```
97
+
98
+ ---
99
+
100
+ ## Completion Report Template
101
+
102
+ Posted to issue at CLOSE. score-issue.sh generates the scorecard section.
103
+
104
+ ```markdown
105
+ ## Completion Report
106
+
107
+ ### Evidence per AC
108
+ - AC-1: → REPLACE:evidence_type REPLACE:evidence_detail
109
+ - AC-2: → REPLACE:evidence_type REPLACE:evidence_detail
110
+
111
+ ### Ship Scorecard
112
+ | Step | Score | Detail |
113
+ |------|-------|--------|
114
+ | SCOPE | REPLACE:score | REPLACE:detail |
115
+ | BUILD | REPLACE:score | REPLACE:detail |
116
+ | VERIFY | REPLACE:score | REPLACE:detail |
117
+ | DURABILITY | REPLACE:score | REPLACE:detail |
118
+ | CLOSE | REPLACE:score | REPLACE:detail |
119
+
120
+ | Metric | Value |
121
+ |--------|-------|
122
+ | First-pass rate | REPLACE:percentage |
123
+ | Iterations | REPLACE:count |
124
+ | Sizing accuracy | REPLACE:match_or_mismatch |
125
+ | Time | REPLACE:hours |
126
+ | Grade | REPLACE:letter_grade |
127
+
128
+ ### Files changed
129
+ - REPLACE:file_line_description
130
+
131
+ ### Follow-ups created
132
+ - REPLACE:issue_refs_or_NONE
133
+ ```
134
+
135
+ ---
136
+
137
+ ## DISCOVERY Comment Template
138
+
139
+ Posted to issue during SCOPE. scope-gate validates this exists.
140
+
141
+ ```markdown
142
+ ## DISCOVERY
143
+
144
+ ### Docs read
145
+ - REPLACE:doc_name (modified: REPLACE:date) — REPLACE:key_constraints
146
+ - REPLACE:doc_name (modified: REPLACE:date) — REPLACE:relevant_modules
147
+
148
+ ### Source spec
149
+ - REPLACE:path_and_section_or_none
150
+
151
+ ### Issue context
152
+ - Related: REPLACE:issue_references
153
+ - Checkpoint: REPLACE:resuming_or_fresh
154
+
155
+ ### Research decision
156
+ - REPLACE:unknowns_or_no_unknowns
157
+
158
+ ### Constraints
159
+ - REPLACE:constraint_with_citation
160
+ ```
161
+
162
+ ---
163
+
164
+ ## Anti-Gaming Table Template
165
+
166
+ Posted to issue during SCOPE after writing ACs. scope-gate validates defenses exist. DA creates this manually or via quick adversarial agent -- gate validates, doesn't create.
167
+
168
+ ~~~markdown
169
+ ## Anti-Gaming
170
+
171
+ | # | Shortcut | Why It Games | Defense Command | Expected |
172
+ |---|---|---|---|---|
173
+ | AG-1 | REPLACE:shortcut_description | REPLACE:why_it_games | REPLACE:defense_command | REPLACE:expected_result |
174
+ | AG-2 | REPLACE:shortcut_description | REPLACE:why_it_games | REPLACE:defense_command | REPLACE:expected_result |
175
+ ~~~
176
+
177
+ > **Defense Command** should use a known ag-test-runner pattern or a raw shell command:
178
+ >
179
+ > Known patterns (run via `bash skills/ship/ag-test-runner.sh --type PATTERN ...`):
180
+ > - `GREP_CHECK --file FILE --pattern PAT --op (eq|gte|lte) --count N` — count grep matches
181
+ > - `CURL_API --url URL --jq FILTER --op (eq|gte|contains) --expected VAL` — check API response
182
+ > - `FILE_EXISTS --path PATH` — verify file/glob exists
183
+ > - `TEST_RUN --command CMD` — run command, check exit 0
184
+ > - `HARDCODE_CHECK --file FILE --pattern PAT` — verify pattern has 0 matches
185
+ > - `CHANGE_DETECT --add FILE --endpoint URL --jq FIELD` — verify endpoint detects file changes
186
+ > - `THRESHOLD_CHECK --command CMD --op OP --value N` — verify numeric output meets threshold
187
+ >
188
+ > Raw shell commands (grep, curl, ls, bun, etc.) are also accepted. verify-gate Check 10 executes these at VERIFY time.
189
+
190
+ > **Prompt for adversarial agent (optional):** "Given this goal statement and these ACs, list every way someone could claim success without actually delivering the goal. For each shortcut, propose a mechanical defense command. You must NOT see the implementation code -- attack from the 'lazy implementer' perspective."
191
+
192
+ ---
193
+
194
+ ## Checkpoint Compaction Template
195
+
196
+ Required fields for context recovery. iteration-gate validates these exist.
197
+
198
+ ```markdown
199
+ ## Current State
200
+ - Phase: [SCOPE | BUILD | VERIFY | ITERATION]
201
+ - ACs: N/M passed (AC-1 ✓, AC-2 ✗, AC-3 pending)
202
+ - Last failure: [AC-N — what failed and why]
203
+ - Iterations: N (Tier 1 at 3, Tier 2 at 6)
204
+
205
+ ## Next Action
206
+ [Exact next step — not vague, not a plan, one specific action]
207
+
208
+ ## Files Changed
209
+ - [file:line — what changed]
210
+
211
+ ## Gate Status
212
+ - scope-gate: [PASS | FAIL (which check)]
213
+ - verify-gate: [PASS | FAIL | not yet run]
214
+ - durability-gate: [PASS | FAIL | not yet run]
215
+ ```
@@ -0,0 +1,68 @@
1
+ #!/usr/bin/env bash
2
+ # batch-status.sh — Track issue status during AFK batch execution
3
+ # Usage:
4
+ # bash batch-status.sh init <batch-id>
5
+ # bash batch-status.sh update <batch-id> <issue> <phase> <status>
6
+ # bash batch-status.sh report <batch-id>
7
+ set -euo pipefail
8
+
9
+ RUNGATE_DIR="${RUNGATE_WORK_DIR:-$HOME/.rungate}"
10
+ ACTION="${1:?Usage: batch-status.sh <init|update|report> ...}"
11
+ BATCH_ID="${2:?Usage: batch-status.sh $ACTION <batch-id> ...}"
12
+ STATUS_FILE="$RUNGATE_DIR/batch-$BATCH_ID/batch-status.json"
13
+
14
+ case "$ACTION" in
15
+ init)
16
+ mkdir -p "$RUNGATE_DIR/batch-$BATCH_ID"
17
+ echo '{"batchId":"'"$BATCH_ID"'","startedAt":"'"$(date -u +%Y-%m-%dT%H:%M:%SZ)"'","issues":[]}' > "$STATUS_FILE"
18
+ echo "PASS: Initialized batch $BATCH_ID at $STATUS_FILE"
19
+ ;;
20
+ update)
21
+ ISSUE="${3:?Usage: batch-status.sh update <batch-id> <issue> <phase> <status>}"
22
+ PHASE="${4:?Usage: batch-status.sh update <batch-id> <issue> <phase> <status>}"
23
+ STATUS="${5:?Usage: batch-status.sh update <batch-id> <issue> <phase> <status>}"
24
+
25
+ if [[ ! -f "$STATUS_FILE" ]]; then
26
+ echo "FAIL: batch $BATCH_ID not initialized" >&2; exit 1
27
+ fi
28
+
29
+ TMP_FILE="$STATUS_FILE.tmp.$$"
30
+ python3 -c "
31
+ import json, sys
32
+ data = json.load(open('$STATUS_FILE'))
33
+ issues = data.get('issues', [])
34
+ updated = False
35
+ for i in issues:
36
+ if i['issue'] == $ISSUE:
37
+ i['phase'] = '$PHASE'
38
+ i['status'] = '$STATUS'
39
+ i['updatedAt'] = '$(date -u +%Y-%m-%dT%H:%M:%SZ)'
40
+ updated = True
41
+ break
42
+ if not updated:
43
+ issues.append({'issue': $ISSUE, 'phase': '$PHASE', 'status': '$STATUS', 'updatedAt': '$(date -u +%Y-%m-%dT%H:%M:%SZ)'})
44
+ data['issues'] = issues
45
+ json.dump(data, open('$TMP_FILE', 'w'), indent=2)
46
+ "
47
+ mv "$TMP_FILE" "$STATUS_FILE"
48
+ echo "PASS: Updated issue $ISSUE → $PHASE/$STATUS"
49
+ ;;
50
+ report)
51
+ if [[ ! -f "$STATUS_FILE" ]]; then
52
+ echo "FAIL: batch $BATCH_ID not initialized" >&2; exit 1
53
+ fi
54
+ python3 -c "
55
+ import json
56
+ data = json.load(open('$STATUS_FILE'))
57
+ print(f'Batch: {data[\"batchId\"]} (started {data[\"startedAt\"]})')
58
+ for i in data.get('issues', []):
59
+ print(f' #{i[\"issue\"]}: {i[\"phase\"]}/{i[\"status\"]} (updated {i.get(\"updatedAt\", \"?\")})')
60
+ total = len(data.get('issues', []))
61
+ done = len([i for i in data.get('issues', []) if i['status'] == 'DONE'])
62
+ print(f'Progress: {done}/{total}')
63
+ "
64
+ ;;
65
+ *)
66
+ echo "FAIL: unknown action '$ACTION' — use init|update|report" >&2; exit 1
67
+ ;;
68
+ esac
@@ -0,0 +1,67 @@
1
+ #!/usr/bin/env bash
2
+ # chain-compat-check.sh — L3 cross-skill compatibility checking
3
+ # Verifies output of skill N matches input expectations of skill N+1
4
+ # Usage: chain-compat-check.sh [chain-yaml]
5
+ set -euo pipefail
6
+
7
+ REPO_ROOT="$(cd "$(dirname "$0")/.." && pwd)"
8
+
9
+ echo "=== L3 Chain Compatibility Check ==="
10
+
11
+ # Format: upstream_skill:upstream_artifact:downstream_skill:expected_field
12
+ HANDOFFS=(
13
+ "goal:goal-record.json:ship:successCriteria"
14
+ "goal:goal-record.json:prove:successCriteria"
15
+ "ship:ship-evidence.json:prove:mergeCommitSha"
16
+ "prove:prove-evidence.json:release:verdict"
17
+ "prove:prove-evidence.json:release:commitSHA"
18
+ )
19
+
20
+ PASSES=0
21
+ FAILS=0
22
+
23
+ for handoff in "${HANDOFFS[@]}"; do
24
+ IFS=':' read -r upstream artifact downstream field <<< "$handoff"
25
+
26
+ SCHEMA_FILE=""
27
+ case "$artifact" in
28
+ goal-record.json) SCHEMA_FILE="$REPO_ROOT/schemas/goal-record.schema.json" ;;
29
+ ship-evidence.json) SCHEMA_FILE="$REPO_ROOT/schemas/ship-evidence.schema.json" ;;
30
+ prove-evidence.json) SCHEMA_FILE="$REPO_ROOT/skills/prove/prove-evidence.schema.json" ;;
31
+ release-proof.json) SCHEMA_FILE="$REPO_ROOT/schemas/release-proof.schema.json" ;;
32
+ esac
33
+
34
+ if [[ ! -f "$SCHEMA_FILE" ]]; then
35
+ echo " FAIL ${upstream}->${downstream}: schema $SCHEMA_FILE not found"
36
+ FAILS=$((FAILS + 1))
37
+ continue
38
+ fi
39
+
40
+ HAS_FIELD=$(python3 -c "
41
+ import json
42
+ schema = json.load(open('$SCHEMA_FILE'))
43
+ props = schema.get('properties', {})
44
+ if '$field' in props:
45
+ print('YES')
46
+ else:
47
+ for prop_name, prop_def in props.items():
48
+ if isinstance(prop_def, dict) and 'properties' in prop_def:
49
+ if '$field' in prop_def['properties']:
50
+ print('YES')
51
+ break
52
+ else:
53
+ print('NO')
54
+ " 2>/dev/null || echo "ERROR")
55
+
56
+ if [[ "$HAS_FIELD" == "YES" ]]; then
57
+ echo " PASS ${upstream}->${downstream}: '$field' exists in $artifact schema"
58
+ PASSES=$((PASSES + 1))
59
+ else
60
+ echo " FAIL ${upstream}->${downstream}: '$field' NOT in $artifact schema"
61
+ FAILS=$((FAILS + 1))
62
+ fi
63
+ done
64
+
65
+ echo ""
66
+ echo "=== Compatibility: $PASSES pass, $FAILS fail ==="
67
+ [[ $FAILS -eq 0 ]] && exit 0 || exit 1
@@ -0,0 +1,314 @@
1
+ #!/usr/bin/env bash
2
+ # context-audit.sh — Pipeline transparency for context injection
3
+ # Usage: bash ~/.claude/scripts/context-audit.sh
4
+ # or via skill: /context-audit
5
+ # Rewritten for #147: surfaces filter stages, domain balance, precision trend
6
+
7
+ set -euo pipefail
8
+
9
+ CLAUDE_DIR="${HOME}/.claude"
10
+ STATE_DIR="${CLAUDE_DIR}/MEMORY/STATE"
11
+ AGENTGRIT_STATE="${HOME}/.agentgrit/state"
12
+ SESSION_CTX="${STATE_DIR}/session-context.json"
13
+ AGENTGRIT_CTX="${AGENTGRIT_STATE}/session-context.json"
14
+ RULE_DOMAINS="${CLAUDE_DIR}/MEMORY/LEARNING/STATE/rule-domains.json"
15
+ RECALL_SCORES="${CLAUDE_DIR}/MEMORY/LEARNING/STATE/recall-scores.json"
16
+ RULE_STATS="${AGENTGRIT_STATE}/rule-stats.json"
17
+
18
+ echo ""
19
+ echo "════════════════════════════════════════════════"
20
+ echo " CONTEXT AUDIT"
21
+ echo "════════════════════════════════════════════════"
22
+ echo ""
23
+
24
+ # ── 1. SESSION ──
25
+ if [[ -f "$SESSION_CTX" ]]; then
26
+ python3 -c "
27
+ import json
28
+ from datetime import datetime, timezone
29
+
30
+ with open('$SESSION_CTX') as f:
31
+ ctx = json.load(f)
32
+
33
+ ts = ctx.get('timestamp','')
34
+ domains = ctx.get('domains', [])
35
+ source = ctx.get('domain_source', '?')
36
+ signals = ctx.get('detection_signals', [])
37
+
38
+ try:
39
+ created = datetime.fromisoformat(ts.replace('Z', '+00:00'))
40
+ age_h = (datetime.now(timezone.utc) - created).total_seconds() / 3600
41
+ expired = age_h > 24
42
+ except:
43
+ age_h = 0
44
+ expired = False
45
+
46
+ print('-- SESSION --')
47
+ print(f' Created: {ts[:19]}')
48
+ print(f' Age: {age_h:.1f}h {\"!! EXPIRED\" if expired else \"\"}')
49
+ print(f' Domains: {len(domains)} -- {\", \".join(domains)}')
50
+ print(f' Source: {source}')
51
+ print(f' Signals: {\", \".join(signals) if signals else \"none\"}')
52
+ print()
53
+ "
54
+ else
55
+ echo "-- SESSION --"
56
+ echo " No session-context.json found at ${SESSION_CTX}"
57
+ echo ""
58
+ fi
59
+
60
+ # ── 2. PIPELINE ──
61
+ if [[ -f "$SESSION_CTX" ]]; then
62
+ python3 -c "
63
+ import json
64
+
65
+ with open('$SESSION_CTX') as f:
66
+ ctx = json.load(f)
67
+
68
+ pipeline = ctx.get('pipeline', {})
69
+ if not pipeline:
70
+ print('-- PIPELINE --')
71
+ print(' No pipeline metadata (session predates #147)')
72
+ print()
73
+ else:
74
+ raw_clusters = pipeline.get('rawCandidateCount', 0)
75
+ raw_ag = pipeline.get('rawAgentgritCount', 0)
76
+ suppressed = pipeline.get('suppressed', [])
77
+ diversity = pipeline.get('diversityDropped', [])
78
+ bm25 = pipeline.get('bm25Stats', {})
79
+ final_rules = ctx.get('loaded_rule_ids', [])
80
+ bm25_total = bm25.get('total', 0)
81
+ bm25_passed = bm25.get('passed', 0)
82
+
83
+ sup_count = len(suppressed)
84
+ div_count = len(diversity)
85
+
86
+ print('-- PIPELINE --')
87
+ print(f' Graph clusters: {raw_clusters} raw')
88
+ print(f' AgentGrit rules: {raw_ag} raw')
89
+ print(f' Always-on suppressed: -{sup_count:>3} (from both sources)')
90
+ print(f' Diversity enforced: -{div_count:>3} (cap {3}/domain per-source)')
91
+ print(f' BM25 (learned): {bm25_total} total -> {bm25_passed} passed')
92
+ print(f' Final injected: {len(final_rules)}')
93
+ print(f' Sources: clusters + agentgrit + learned')
94
+ print()
95
+
96
+ if suppressed:
97
+ print(' Suppressed IDs:')
98
+ for rid in suppressed:
99
+ print(f' - {rid}')
100
+ print()
101
+
102
+ if diversity:
103
+ print(' Diversity-dropped IDs:')
104
+ for rid in diversity:
105
+ print(f' - {rid}')
106
+ print()
107
+ "
108
+ fi
109
+
110
+ # ── 3. DOMAIN BALANCE ──
111
+ if [[ -f "$SESSION_CTX" ]]; then
112
+ python3 -c "
113
+ import json
114
+
115
+ with open('$SESSION_CTX') as f:
116
+ ctx = json.load(f)
117
+
118
+ pipeline = ctx.get('pipeline', {})
119
+ dist = pipeline.get('domainDistribution', {})
120
+ domains = ctx.get('domains', [])
121
+
122
+ if not dist:
123
+ # Fall back: compute from loaded_rule_ids + rule-domains.json
124
+ try:
125
+ with open('$RULE_DOMAINS') as f:
126
+ rd = json.load(f)
127
+ rules_meta = rd.get('rules', {})
128
+ dist = {}
129
+ for rid in ctx.get('loaded_rule_ids', []):
130
+ d = rules_meta.get(rid, {}).get('domains', ['unknown'])[0]
131
+ dist[d] = dist.get(d, 0) + 1
132
+ except:
133
+ pass
134
+
135
+ if dist:
136
+ print('-- DOMAIN BALANCE --')
137
+ max_count = max(dist.values()) if dist else 0
138
+ cap = 3 # MAX_PER_DOMAIN from GraphContext.hook.ts
139
+ sorted_domains = sorted(dist.items(), key=lambda x: -x[1])
140
+ for domain, count in sorted_domains:
141
+ bar = '#' * count
142
+ over = ' !! >3 combined (cap is per-source)' if count > cap else ''
143
+ match = ' (detected)' if domain in domains else ''
144
+ print(f' {domain:<20} {bar:<15} {count}{over}{match}')
145
+ print()
146
+ "
147
+ fi
148
+
149
+ # ── 4. INJECTED RULES (with relevance) ──
150
+ if [[ -f "$SESSION_CTX" ]]; then
151
+ python3 -c "
152
+ import json, os
153
+
154
+ with open('$SESSION_CTX') as f:
155
+ ctx = json.load(f)
156
+
157
+ # Load rule domains for relevance check
158
+ rule_domains = {}
159
+ try:
160
+ with open('$RULE_DOMAINS') as f:
161
+ rd = json.load(f)
162
+ rule_domains = rd.get('rules', {})
163
+ except:
164
+ pass
165
+
166
+ # Load rule stats for quality metrics
167
+ stats_map = {}
168
+ try:
169
+ with open('$RULE_STATS') as f:
170
+ stats = json.load(f)
171
+ stats_map = {s['ruleId']: s for s in stats}
172
+ except:
173
+ pass
174
+
175
+ session_domains = set(ctx.get('domains', []))
176
+ rules = ctx.get('loaded_rule_ids', [])
177
+
178
+ print('-- INJECTED RULES --')
179
+ print(f' {\"#\":<3} {\"Rule\":<55} {\"Avg\":>5} {\"Sess\":>4} Relevance')
180
+ print(f' {\"-\"*3} {\"-\"*55} {\"-\"*5} {\"-\"*4} {\"-\"*10}')
181
+
182
+ match_count = 0
183
+ mismatch_count = 0
184
+
185
+ for i, rid in enumerate(rules, 1):
186
+ stat = stats_map.get(rid, {})
187
+ avg = stat.get('avgCorrelatedRating', 0)
188
+ sessions = len(stat.get('sessionRatings', []))
189
+
190
+ name = rid[:55]
191
+
192
+ # Relevance: check if rule's domains overlap with session domains
193
+ rule_meta = rule_domains.get(rid, {})
194
+ rule_doms = set(rule_meta.get('domains', []))
195
+ if rule_doms & session_domains:
196
+ relevance = 'MATCH'
197
+ match_count += 1
198
+ elif not rule_doms:
199
+ relevance = '? no domain'
200
+ mismatch_count += 1
201
+ else:
202
+ relevance = 'MISMATCH'
203
+ mismatch_count += 1
204
+
205
+ print(f' {i:<3} {name:<55} {avg:>5.1f} {sessions:>4} {relevance}')
206
+
207
+ print()
208
+ print(f' Relevance: {match_count} MATCH, {mismatch_count} MISMATCH/unknown')
209
+ print()
210
+ "
211
+ fi
212
+
213
+ # ── 5. PRECISION TREND ──
214
+ if [[ -f "$RECALL_SCORES" ]]; then
215
+ python3 -c "
216
+ import json
217
+
218
+ with open('$RECALL_SCORES') as f:
219
+ scores = json.load(f)
220
+
221
+ p5 = scores.get('mean_precision5', 0)
222
+ target = 0.64
223
+ r5 = scores.get('mean_recall5', 0)
224
+ r10 = scores.get('mean_recall10', 0)
225
+ mrr = scores.get('mean_mrr', 0)
226
+ sessions = scores.get('sessions_evaluated', 0)
227
+ last_int = scores.get('last_intervention', 'unknown')[:10]
228
+
229
+ delta = p5 - target
230
+ direction = '+' if delta >= 0 else ''
231
+
232
+ print('-- PRECISION TREND --')
233
+ print(f' precision@5: {p5:.3f} (target: {target}, delta: {direction}{delta:.3f})')
234
+ print(f' recall@5: {r5:.3f}')
235
+ print(f' recall@10: {r10:.3f}')
236
+ print(f' MRR: {mrr:.3f}')
237
+ print(f' Sessions evaluated:{sessions}')
238
+ print(f' Last intervention: {last_int}')
239
+ print()
240
+ "
241
+ else
242
+ echo "-- PRECISION TREND --"
243
+ echo " No recall-scores.json found"
244
+ echo ""
245
+ fi
246
+
247
+ # ── 6. SUPPRESSED RULES ──
248
+ python3 -c "
249
+ import json
250
+
251
+ # Read suppressed list from pipeline metadata if available
252
+ suppressed_ids = []
253
+ try:
254
+ with open('$SESSION_CTX') as f:
255
+ ctx = json.load(f)
256
+ suppressed_ids = ctx.get('pipeline', {}).get('suppressed', [])
257
+ except:
258
+ pass
259
+
260
+ # Also show the static ALWAYS_ON_SUPPRESSED list for reference
261
+ always_on = [
262
+ 'feedback_capture_decisions_immediately',
263
+ 'feedback_always_update_docs',
264
+ 'feedback_verify_before_answering',
265
+ 'feedback_iterative_quality_loop',
266
+ 'success_real-data-honest-gaps',
267
+ 'feedback_exhaustive-search-before-claiming-nonexistence',
268
+ 'feedback_read-background-task-output',
269
+ 'feedback_avoid-unnecessary-doc-regeneration',
270
+ 'feedback_proactive-spot-check-patterns',
271
+ 'success_technical-explanation-direct-and-structu',
272
+ ]
273
+
274
+ # Load rule names from knowledge graph if available
275
+ rule_names = {}
276
+ try:
277
+ with open('${AGENTGRIT_STATE}/knowledge-graph.json') as f:
278
+ graph = json.load(f)
279
+ for node in graph.get('nodes', []):
280
+ rule_names[node.get('id', '')] = node.get('text', '')[:60]
281
+ except:
282
+ pass
283
+
284
+ print('-- SUPPRESSED RULES --')
285
+ print(f' Always-on suppression list ({len(always_on)} rules):')
286
+ for rid in always_on:
287
+ hit = ' (was in candidates)' if rid in suppressed_ids else ''
288
+ name = rule_names.get(rid, '')
289
+ label = f' -- {name}' if name else ''
290
+ print(f' {rid}{label}{hit}')
291
+ print()
292
+ "
293
+
294
+ # ── RECENT SESSIONS ──
295
+ if [[ -f "${AGENTGRIT_STATE}/session-context-history.jsonl" ]]; then
296
+ echo "-- RECENT SESSIONS --"
297
+ tail -5 "${AGENTGRIT_STATE}/session-context-history.jsonl" | python3 -c "
298
+ import sys, json
299
+ for line in sys.stdin:
300
+ line = line.strip()
301
+ if not line: continue
302
+ try:
303
+ d = json.loads(line)
304
+ ts = d.get('timestamp','')[:16]
305
+ rules = d.get('rulesInjectedCount', '?')
306
+ kb = d.get('rulesInjectedKB', '?')
307
+ domains = len(d.get('domains', []))
308
+ print(f' {ts} | {rules} rules | {kb} KB | {domains} domains')
309
+ except: pass
310
+ "
311
+ echo ""
312
+ fi
313
+
314
+ echo "════════════════════════════════════════════════"