Compile enterprise policies into project-level CLAUDE.md files. Use when onboarding a new project (Phase 1) or when the user asks about policy requirements, compliance, or CLAUDE.md generation.
Generates project-level artifacts from enterprise policy templates:
Ask the developer: "What type of data will this agent handle?"
| Answer | Tier | Policies |
|---|---|---|
| PII, financial, health data |
| Tier 1 |
| All 4 policies |
| Internal business data | Tier 2 | Security + Architecture + Testing |
| Public/non-sensitive | Tier 3 | Security basics only |
Located in workspace/policies/:
security-standards.md (v1) — All tiersagent-architecture.md (v1) — Tier 1-2compliance-requirements.md (v1) — Tier 1 onlytesting-standards.md (v1) — Tier 1-2<!-- plato-policy: security-v1[, architecture-v1][, compliance-v1][, testing-v1] -->
<!-- plato-tier: {1|2|3} -->
<!-- plato-generated: {date} -->
# {Project Name} — Development Guide
## Project Overview
{From kickoff conversation}
## Architecture Requirements
{From agent-architecture.md, if tier 1-2}
## Security Requirements
{From security-standards.md, always included}
## Compliance Requirements
{From compliance-requirements.md, if tier 1}
## Testing Requirements
{From testing-standards.md, if tier 1-2}
## Readiness Checklist
{Generated from applicable policies}
Include policy versions in HTML comments at the top of CLAUDE.md. This enables drift detection: Plato can scan repos and compare the embedded version against the current policy version.
See workspace/examples/acme-corp/CLAUDE.md for a complete Tier 1 example
(fintech customer support agent with PII handling and refund approval workflow).