| name | conversational-agent |
| description | Playbook for translating natural language requests about tasks, profiles, locking, and validation into CLI helper executions. |
Conversational Agent Playbook
When the user asks to switch profiles, manage tasks, lock modules, or audit code in natural language:
1. Profile Rotation & Matching
2. Lock Management
- User says: "Lock validate" or "Unlock git_api"
- Action:
- For locking:
./helper.sh lock <module-name>
- For unlocking:
./helper.sh lock <module-name> --release
3. Issue and Task Tracking
- User says: "Create task for web UI" or "Mark issue-123 as complete"
- Action:
- For creating issues:
./helper.sh issue create <id> "<title>"
- For closing issues:
./helper.sh issue close <id>
- For listing tasks:
./helper.sh issue list
4. Run Audits & Checks
- User says: "Check my workspace health" or "Validate project before commit"
- Action:
- For general diagnostics:
./helper.sh doctor
- For validation audits:
./helper.sh validate
5. Local Visual Dashboard
- User says: "Show dashboard", "Open web UI", or "Run local status dashboard"
- Action:
- Run:
./helper.sh dashboard
- This will start a local HTTP server and open the interactive dashboard in a browser.
6. Human Validation Bypass
- User says: "Bypass validation" or "Skip audit checks"
- Action:
- Run the command with
--bypass or with the environment variable AAC_BYPASS_COMPLIANCE=1 prefix.
- Example:
AAC_BYPASS_COMPLIANCE=1 git commit -m "feat(scope): descriptive message"
- Or:
./helper.sh validate --bypass