TCA Audit

Executes the component audit phase (Phase 1) of a Total Codebase Audit. Manages sub-agent launches, progress tracking, context management, and commit discipline.

For reference:

intent/docs/total-codebase-audit.md

Procedure

1. Check progress

Run the progress script to see current state:

bash "$(find ~/.claude/skills/in-tca-audit -name tca-progress.sh 2>/dev/null | head -1)" \
  --tca-dir intent/st/STXXXX

This shows which WPs are complete, pending, or in-progress, with violation counts.

2. Select next WP batch

From the batch ordering in

design.md

3. Pre-flight checklist (per WP)

Before launching each sub-agent:

1. Run

   /compact

   to reclaim context window space
2. Verify file manifest -- confirm all files in the WP's

   info.md

   exist:

   for f in $(grep "^- \`" WP/NN/info.md | sed 's/.*`\(.*\)`.*/\1/'); do
     [ -f "$f" ] || echo "MISSING: $f"
   done
   

3. Check context usage -- if above 70%, consider fresh session
4. Verify previous WP committed -- no uncommitted audit files

4. Launch sub-agent

Use the prompt template with these required elements:

You are performing a forensic code audit of the **{Component Name}** subsystem
in the {Project} application.

### Audit Rules

| #   | Rule                    | What to Check                              |
| --- | ----------------------- | ------------------------------------------ |
| R1  | {rule name}             | {brief description of what to look for}    |
...

### Files to Audit

1. `{path/to/file1}`
2. `{path/to/file2}`
...

### Special Focus

{R-numbers and why they matter for this component}

### Cross-WP Highlander Check

- vs. WP-XX: {what logic might be duplicated and why}
- vs. WP-YY: {what logic might be duplicated and why}

### Instructions

Read EVERY file listed above. For each file, check every function against
all rules. Report violations in this exact format:

#### V{N}: {short title}

- **File**: `{path}:{line(s)}`
- **Rule**: R{N} -- {rule name}
- **Severity**: High | Medium | Low
- **Confidence**: HIGH | MEDIUM | LOW
- **Description**: {what the violation is}
- **Remedy**: {how to fix it}

After all violations, add a summary table:

### Summary

| Severity  | Count |
| --------- | ----- |
| High      | X     |
| Medium    | Y     |
| Low       | Z     |
| **Total** | **T** |

Write the complete audit to `{path/to/WP/NN/socrates.md}`.

Be thorough. Do NOT skip files. Do NOT invent violations -- only report
what you actually see in the code. If a file has no violations, say so
explicitly.

Agent selection:

diogenes

general-purpose

Explore

elixir

The prompt is the quality driver, not the agent type.

5. Post-WP

After each sub-agent completes:

1. Record metadata at the top of

   WP/{NN}/socrates.md

   as the first non-heading line (or right after the H1 title). Format:

   **Agent**: {type}; **Turns**: N; **Raw hits**: N; **FPs**: N

   . Example:

   **Agent**: elixir; **Turns**: 24; **Raw hits**: 9; **FPs**: 2

   . This line is queryable across audits and feeds directly into the "Sub-Agent Effectiveness" section of the final feedback report. Without it, cross-TCA comparisons have to record "not tracked" for agent effectiveness, which is exactly what happened on the Lamplight ST0121 run that motivated this rule.
2. Commit immediately:

   git add WP/{NN}/socrates.md && git commit -m "audit: WP-{NN} {component}"

3. Log the summary in your running tally
4. Run progress check to update status

6. Repeat

Continue launching sub-agents for each WP in the current batch, then move to the next batch per the ordering in design.md.

7. Final check

After all component WPs are done, run the progress script one final time:

bash "$(find ~/.claude/skills/in-tca-audit -name tca-progress.sh 2>/dev/null | head -1)" \
  --tca-dir intent/st/STXXXX

Exit code 0 confirms all WPs complete. Proceed to

/in-tca-synthesize

Crash Prevention

/compact

Important Notes

• Never batch commits -- commit after every single WP
• The anti-hallucination instruction is not optional
• Include the confidence field in every violation
• If context usage exceeds 70%, start a fresh session rather than risk truncation