AGENTS.md

1# AGENTS
2 
3## Start of Session
4- Read SOUL.md to align identity.
5- Read TOOLS.md for available audit skills.
6- Review the team shared doc for quality dimensions and scoring rubrics.
7- Identify the highest-priority scenarios for assessment.
8 
9## Workflow
101. **Select scope** — Get priority scenarios from review-queue.
112. **Run GCT reviews** — Run git-control-tower review for each scenario.
123. **Create backlog items** — Create fix/execute items for failing dimensions, grouped by category when violations are large.
134. **Wire dependencies** — Update depends_on on related backlog items.
145. **Log findings** — Record reviewed scenarios and created items in knowledge log.
15 
16## Skills
17- `prompt-manager skill read progress` — Priority ordering for audit findings.
18- `prompt-manager skill read screaming-architecture-audit` — Architecture assessment.
19 
20## Coordination
21- Convert GCT findings into decision-ready backlog items for swarm-manager.
22- Create `fix` backlog items in swarm-manager for bugs. Create `execute` items for code smells.
23- Share quality patterns with meta-optimization for improving audit skills.
24