Skip to main content

​
Story

What if you could show someone exactly how AI agent governance works — not with a slide deck, but with 15 minutes at a terminal? ATHENA is a documentation council: 5 AI agents that draft, review, and publish technical documentation. SCRIBE-01 writes. CURATOR-02 reviews (and cannot be bypassed — the system enforces this cryptographically). EDITOR-04 holds final publication authority. Every step is gated, fingerprinted, and auditable. The ATHENA Showcase is a guided CLI walkthrough that runs the full pipeline live:
  1. Validate — Schema checks on every prompt artifact
  2. Fingerprint — Blake3 content-addressing for tamper detection
  3. Compile — One YAML chain compiles to TypeScript, n8n, and LangGraph
  4. Execute — Run the chain with real LLM calls, pausing at each gate for human approval
  5. Break it — Try to tamper a fingerprint, remove the required reviewer, downgrade a gate. The pipeline catches every violation.

​
Meta Layer

This isn’t a demo of documentation tooling. It’s a demo of governance infrastructure for AI agents — the kind of thing that makes compliance teams stop worrying and engineering teams stop reinventing safety checks. The “break it on purpose” section is what separates STRATT from a template library. When someone tries to remove the mandatory reviewer and the CI pipeline blocks them with FM-04: Protected agent CURATOR-02 missing, they’re seeing what production-grade agent oversight looks like.

​
Signal

The showcase pattern is now replicable across all 15 STRATT councils. ATHENA is the reference implementation. Each council gets: a 3-step chain, 3 sabotage fixtures, a walkthrough document, and a Meridian visualisation page. The onboarding story writes itself — pick a domain (docs, ops, security, finance), run the showcase, see the governance in action. Try it: docs/demo/athena-showcase.md or stratt.dev/demo/athena/