Automated health checks for framework artifacts modified during a pipeline. Validates hook syntax, security patterns (SE-01/SE-02), skill wiring, agent consistency, and routing correctness. Invoked as the final pipeline step when framework artifacts were created, modified, or deleted.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
A direct command skips the review prompt. Inspect the source before running it.
Automated health checks for framework artifacts modified during a pipeline. Validates hook syntax, security patterns (SE-01/SE-02), skill wiring, agent consistency, and routing correctness. Invoked as the final pipeline step when framework artifacts were created, modified, or deleted.
version
1.2.0
model
sonnet
category
validation
invoked_by
both
user_invocable
true
tools
["Read","Bash","Glob","Grep"]
agents
["qa","developer","architect"]
best_practices
["Run all applicable checks from the check matrix","Use git diff as primary change detection method","Never skip SE-02 or SE-01 scans on hook/tool files","Report findings with severity and remediation steps"]
error_handling
strict
streaming
supported
verified
true
lastVerifiedAt
"2026-02-23T00:00:00.000Z"
source
builtin
trust_score
100
provenance_sha
7cb12aa5e0ac92fe
Proactive Audit
Overview
Automated health checks for framework artifacts that were modified during the current pipeline. This skill fills the gap between reactive verification (tests, lint) and proactive framework-level validation (wiring, syntax, security patterns).
Core principle: Framework artifact changes require the same rigor as code changes. If a skill was created, verify it is wired. If a hook was modified, verify it compiles. If an agent was changed, verify its tool/skill lists are consistent.
When to Invoke
Invoke this skill as the final pipeline step whenever ANY of the following paths were created, modified, or deleted during the session:
.claude/hooks/**/*.cjs
.claude/skills/**/SKILL.md
.claude/agents/**/*.md
.claude/workflows/**/*.md
.claude/schemas/**/*.json
.claude/templates/**/*
.claude/CLAUDE.md
.claude/lib/routing/routing-table.cjs
Invocation:
Skill({ skill: 'proactive-audit' });
Mandatory Skills
Skill
Purpose
When
task-management-protocol
Track audit progress
Always
ripgrep
Fast targeted artifact search
During checks
code-semantic-search
Pattern discovery across artifacts
When investigating
context-compressor
Compress large audit results
When output is large
verification-before-completion
Gate completion on zero CRITICAL
Before marking done
memory-search
Check prior audit patterns
At start
Step 1: Detect Changed Artifacts
Use git diff to identify which framework artifacts changed in this session:
Combine all three lists into a deduplicated set of changed artifact paths.
Step 2: Apply Check Matrix
For each changed artifact, apply the relevant checks from this matrix:
Hook Files (.claude/hooks/**/*.cjs)
Check ID
Check
Command
Severity
H-01
Syntax validity
node --check <file>
CRITICAL
H-02
SE-02: raw JSON.parse without safeParseJSON
grep -n "JSON.parse(" <file> then verify safeParseJSON import
HIGH
H-03
SE-01: shell injection via shell: true
grep -n "shell:\\s*true" <file>
HIGH
H-04
Hook registered in settings.json
grep "<hook-filename>" .claude/settings.json
MEDIUM
H-05
Exit code correctness
Verify try/catch wrapping, exit 0 on non-critical errors
MEDIUM
H-02 detail: If JSON.parse( is found, check if the file also imports safeParseJSON from .claude/lib/utils/safe-json.cjs. If not, flag as HIGH finding. Exclude test files (*.test.cjs).