Agent skill
documentation-cascade-audit
Run a hierarchical documentation audit for BioETL using cascade decomposition across doc domains (architecture, providers, contracts, operations, onboarding), aggregate findings into one prioritized report, and generate an actionable remediation plan. Use when users request large-scale doc audits, stale-doc cleanup, or coordinated doc reconciliation after major refactors/releases.
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/other/documentation-cascade-audit
SKILL.md
Documentation Cascade Audit
Overview
Coordinate a multi-scope documentation audit where each scope is analyzed separately and then merged into a single decision-ready report. Use this skill when a single-pass manual review is too large or error-prone.
Startup Context
Read, in this order:
../../../.ai/memory/agent-memory.md../../../.claude/agents/ORCHESTRATION.md../documentation-audit/SKILL.md../documentation-audit/references/audit-checklist.md../documentation-audit/references/report-template.md
Cascade Workflow
- Build audit shards.
- Create shard scopes by domain:
docs/01-overview*,README.md,mkdocs.yml- architecture/ADR docs
- provider docs
- contracts/schemas docs
- operations/monitoring/runbook docs
- Keep shards non-overlapping when possible.
- Execute shard audits in parallel (conceptually or via subagents if available).
- For each shard, apply the checklist from
documentation-audit. - Capture findings with severity and evidence (
file + lines + command).
- Normalize findings.
- Deduplicate repeated findings across shards.
- Merge equivalent root causes under one canonical finding.
- Keep conflicting findings in a dedicated
Needs Clarificationsection.
- Produce consolidated outputs.
- Consolidated audit report.
- Prioritized remediation backlog (
P1/P2/P3). - Sequenced update plan with effort estimate.
Deliverables
Required artifacts:
reports/plans/<task-id>/cascade-audit-report.mdreports/plans/<task-id>/cascade-remediation-plan.mdreports/plans/<task-id>/cascade-open-questions.md
Quality Gates
- Every high-severity claim has evidence.
- Every proposed doc change maps to a concrete file path.
- RULES/REQUIREMENTS/ADRs consistency is explicitly checked.
- The final plan distinguishes factual drift vs style improvements.
Constraints
MUST:
- Prefer documenting current system behavior over intended behavior.
- Mark unresolved uncertainties as
Requires Manual Review. - Keep architecture statements verifiable against code.
MUST NOT:
- Modify production code in this workflow.
- Delete documentation without explicit user approval.
- Hide contradictions between docs and code.
SHOULD:
- Group edits into small, reviewable change sets.
- Provide before/after snippets for high-impact doc changes.
Handoff Format
Return:
- Status:
Completed | Partially Completed | Blocked - Top findings by severity
- File list for proposed changes
- Open questions requiring user decisions
- Link to
cascade-audit-report.md
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?