MarioCasanovaCF/serenissima-harness

Serenissima Harness

A one-page overview of a file-based coordination substrate that lets several AI agents work on one real codebase in parallel — without stepping on each other, without approving their own work, and leaving an auditable trail behind every decision.

The page (index.html) is static: the blackboard DAG and the evolution ledger are rendered from real snapshots of the harness's own state, with the DAG interactive client-side (plotly.js).

Full spec, source, and evidence trail → github.com/MarioCasanovacf/serenissima-harness