cc-os/plugins/cc-architect/workflows/audit-command.md

status	note	model_tier
legacy	For maintenance of existing commands only. Prefer skills for new functionality.	setup frontmatter description instructions report cleanup main main opus main main haiku

Workflow: Audit Command (LEGACY)

LEGACY WORKFLOW: This workflow is for maintaining existing slash commands. For new functionality, prefer creating skills via /cc-architect (create-skill workflow) instead.

When to use

• Evaluating slash command for quality and compliance
• Checking slash-command-architect standards
• Reviewing description and tool permissions

Inputs required

• command_path - Path to command file (e.g., plugin/commands/my-command.md)

Output

• .claude/plugin-data/slash-command-architect/audits/<command-name>-audit.md (gitignored)
• Summary with verdict (Ready / Not Ready)
• Delete report after user acts on findings

---

Note on scope

Commands are simpler than skills. This audit is lighter-weight with fewer subtasks.

---

Subtasks

1. Pre-audit context loading

Subagent: Main.

Check for:

• invariants.md in command's plugin root
• .decisions/ directory

Record in scratch:

• Applicable invariants (or "none")
• Prior decisions affecting commands (or "none")

---

2. Validate path and setup

Subagent: Main.

• Verify command_path exists, is .md file
• Verify in commands/ directory
• Extract command name

---

3. Frontmatter audit

Subagent: Main (inline).

Checks:

• Valid YAML frontmatter
• Required fields: name, description
• Name kebab-case, matches filename
• If allowed-tools: valid tool names array

---

4. Description compliance audit

Subagent: opus. Read: ${CLAUDE_PLUGIN_ROOT}/references/layout/descriptions.md

Checks:

• Follows "When/Why, Not How"
• No implementation details
• No escape hatches ("if needed", "when appropriate")
• Clear trigger

Write: scratch/description-findings.md

---

5. Instruction completeness audit

Subagent: Main.

Checks:

• Clear context section
• Instructions actionable
• Handles user request interpretation
• Tool permissions match requirements

---

6. Assemble report

Subagent: Main.

Do:

• Combine findings
• Classify by severity (significant/minor/polish)
• Verdict: Ready (0 significant) or Not Ready (1+ significant)
• Write: .claude/plugin-data/slash-command-architect/audits/<command-name>-audit.md

Severity for commands:

• Significant: Invalid frontmatter, missing fields, escape hatches
• Minor: Broad tool permissions, unclear instructions
• Polish: Wording, organization

---

7. Act on findings

Subagent: Main.

• Walk through significant findings first
• Offer to fix or let user handle
• After user confirms: delete report

---

8. Record decisions

Subagent: Main.

When to record:

• Exceptions to description formula
• Unusual tool permission grants

How:

• Template: references/tool-patterns/templates/decision-record-template.md
• Write: <plugin_path>/.decisions/YYYY-MM-DD-<topic>.md

Prompt: "Were any significant decisions made?"