autonomous-rca-governance
Skill40 starsChanged 5 days agoApache-2.0View on GitHub
Use when a core runtime, CI, or UAT release surface is failing and the task is to classify the failure, choose the smallest safe remediation, rerun authoritative checks, and preserve machine-readable RCA artifacts.
In this repository69 other files it gives its agents.
- AGENTS.mdAGENTS.md
- hushh-webapp/AGENTS.mdAGENTS.md
- impeccableSkill
- agent-orchestration-governanceSkill
- analytics-observability-governanceSkill
- animateSkill
- animation-vocabularySkill
- apple-designSkill
- backend-agents-operonsSkill
- backend-api-contractsSkill
- backend-runtime-governanceSkill
- backendSkill
- brandkitSkill
- client-env-paritySkill
- codex-skill-authoringSkill
- comms-communitySkill
- context-refreshSkill
- contributor-onboardingSkill
- docs-governanceSkill
- emil-design-engSkill
- find-animation-opportunitiesSkill
- founder-brief-curationSkill
- frontend-architectureSkill
- frontend-cache-coherenceSkill
- frontend-design-systemSkill
- frontend-native-surface-mapperSkill
- frontendSkill
- frontend-surface-placementSkill
- future-plannerSkill
- github-contribution-governanceSkill
- hushh-consent-mcpSkill
- hushh-icon-themeSkill
- iam-consent-governanceSkill
- imagegen-frontend-mobileSkill
- imagegen-frontend-webSkill
- improve-animationsSkill
- industrial-brutalist-uiSkill
- kai-test-account-resetSkill
- less-text-karma-jiSkill
- mcp-developer-surfaceSkill
- mobile-nativeSkill
- mobile-parity-auditSkill
- mobile-plugin-contractsSkill
- morphy-axSkill
- one-voice-governanceSkill
- oss-license-governanceSkill
- pdf-artifact-generationSkill
- pick-ui-librarySkill
- pkm-upgrade-rehearsalSkill
- planning-boardSkill
- pr-governance-reviewSkill
- profile-uiSkill
- prototypeSkill
- puppy-one-harnessSkill
- quality-contractsSkill
- release-ios-appstoreSkill
- repo-contextSkill
- repo-operationsSkill
- review-animationsSkill
- reviewer-app-testingSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
8 rules. About 0 are common, 2 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 12
Avoid overlap with broad repo discovery or product implementation that is not anchored to a failing authority surface.
- Very specific: line 68
Read unevaluable_checks FIRST, before any of the above.
- Very specific: line 92
Do not call the RCA task complete while a relevant core run for the touched SHA is still queued, in_progress, or requested.
- Distinctive: line 93
Stop only when the blocking classifications are empty and the relevant core run chain is terminal green, or a real permissions/product …
- Very specific: line 98
Use repo-operations when the fix is primarily in deploy workflows, Cloud Run, or Secret Manager wiring.
- Very specific: line 99
Use backend-runtime-governance when the fix is in backend runtime ownership or service behavior.
- Very specific: line 100
Use quality-contracts when the authoritative proof is missing or the semantic gate itself is wrong.
- Very specific: line 101
Use contributor-onboarding when bootstrap/doctor/runtime files cause false RCA signals.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Hussh Autonomous RCA Governance Skill
Purpose and Trigger
- Primary scope:
autonomous-rca-governance-intake - Trigger on core runtime, CI, deploy, and semantic-release failures that need repeated RCA and a structured remediation loop.
- Avoid overlap with broad repo discovery or product implementation that is not anchored to a failing authority surface.
Coverage and Ownership
- Role:
owner - Owner family:
autonomous-rca-governance
Owned repo surfaces:
.agents/skills/autonomous-rca-governancescripts/ops/verify-env-secrets-parity.pyscripts/ops/verify_uat_release.pyscripts/ci/verify-runtime-config-contract.pyscripts/ci/cloudrun-rollback.sh
Non-owned surfaces:
repo-operationsbackend-runtime-governancequality-contractscontributor-onboarding
Do Use
- Distinguishing
secret_missingvsruntime_mount_legacyvsruntime_behavior_failed, and any of those fromcheck_unevaluable(the check never ran, so nothing is known). - Running repeated RCA for UAT, core runtime, or CI failures with machine-readable artifacts.
- Tightening the repo-local RCA taxonomy, resume artifacts, or canonical RCA command surface.
Do Not Use
- Broad feature work that is not anchored to a failing core authority surface.
- Docs-home or taxonomy work that does not change the RCA loop itself.
- Production promotion decisions.
Read First
docs/operations/releases/ci.mddocs/operations/configuration/env-and-secrets.md.agents/skills/repo-operations/SKILL.md.agents/skills/quality-contracts/SKILL.md
Workflow
- Start with
./bin/hushh agents rca --surface <uat|runtime|ci>and preserve the generated report. CI observes the current branch's existing PR by default; use paired--ci-pr <number> --ci-sha <full-sha>for a frozen candidate. It never starts a suite on missing or refused hosted evidence.--run-local-ciis an explicit local execution request, not a monitoring step. - Classify failures before editing anything. Domain classes, each meaning
"a check RAN and found this":
secret_missingruntime_mount_missingruntime_mount_legacyruntime_behavior_failedsemantic_verifier_failedsmoke_overlay_dependency_leakdb_contract_driftruntime_contract_driftcore_ci_failed
- Read
unevaluable_checksFIRST, before any of the above. It is reported separately and is deliberately not a domain class:check_unevaluablemeans this runner could not ask the question, so nothing was verified either way. Each entry names the check, the reason, and the domain class it would otherwise have been blamed for.- Status is three-state:
healthy,blocked,unevaluable. Exit codes are0,1,2respectively.can_push_branchis false for the last two, because in neither case has anything been verified. - Never remediate a domain class that came from an unevaluable check. On
2026-08-29 the uat surface reported
db_contract_driftand "Resolve DB release-contract drift before treating the surface as deployable" becauseverify_runtime_db_contract.shdied atimport asyncpg. The contract was fine. In the same run the semantic verifier died atimport dotenvand produced no classification at all, so a crashed release check read as a clean one. Two opposite lies, six lines apart in one function. - The deploy gates model this correctly and always did:
.github/workflows/deploy-uat.ymlseparatessemantic_verifier_failedfromruntime_behavior_failedand testsdb_outcome == "failure"rather than!= "success", so a skipped step never becomes drift. Keep the local runner and the gate in agreement; the divergence is what let the local one lie.
- Status is three-state:
- Apply the smallest safe remediation for the highest-signal blocking class first. For provider failures, use the Admin SOP's execution-bound diagnosis.
- Run the smallest affected authoritative check after remediation, including a negative control for a real security regression. Reuse unchanged exact evidence. Run the final core mirror once; do not repeat a full local suite while hosted validation runs. Repeat only to investigate a concrete regression or after an environment repair.
- Do not call the RCA task complete while a relevant core run for the touched SHA is still
queued,in_progress, orrequested; keep monitoring until GitHub reaches a terminal state. - Stop only when the blocking classifications are empty and the relevant core run chain is terminal green, or a real permissions/product boundary is explicit.
- Advisory doc/skill drift should be recorded, but it must not block the loop unless it masks a core runtime failure.
Handoff Rules
- Use
repo-operationswhen the fix is primarily in deploy workflows, Cloud Run, or Secret Manager wiring. - Use
backend-runtime-governancewhen the fix is in backend runtime ownership or service behavior. - Use
quality-contractswhen the authoritative proof is missing or the semantic gate itself is wrong. - Use
contributor-onboardingwhen bootstrap/doctor/runtime files cause false RCA signals.
Required Checks
Select only the affected surface; these are alternatives, not a three-surface sweep.
./bin/hushh agents rca --surface ci --text
python3 -m py_compile .agents/skills/autonomous-rca-governance/scripts/rca_runner.pyDoes it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

