Slipmark is live and open source

Autonomous agents now produce clearance verdicts: security audits, QA passes, review approvals. Nothing usually ties a verdict to what was actually inspected. Slipmark does. When an agent approves, its verdict is recorded against a SHA-256 hash of the files it read. Before anything ships, the gate hashes those files again. If anything has changed, it refuses, so nothing ships against a stale verdict.

It also includes drift diagnosis for AI features. In ReplyOS we use it to lock behaviours we never want to change silently, including tenant isolation, enquiry capture and email threading.

One security audit gate in our own nightly pipeline of more than 20 agents inspected the wrong branch from 20 May and reported green for months. When we pointed the first version of Slipmark at the 23 agent pull requests open at the time, it refused all 23. We have written up the full story separately.

A slip mark is the paint line engineers draw across a bolt once it has been tightened and checked. If the bolt moves, the line breaks. The rename from DriftGuard also gives the project a name of its own.

Slipmark runs as a GitHub Action. Add it to a workflow with one line: uses: founder-ship-it/Slipmark@v0.2.0. The source, documentation and issues are on GitHub, and the Action is listed on the GitHub Marketplace as Slipmark Verify.

Slipmark is free and stays open source. A paid layer is planned: a hosted dashboard, evidence reports, team policies and setup help. You can leave an email address on the Slipmark page to hear when it is ready.

What it does not do: Slipmark does not prove that an agent behaves correctly. It detects when what was approved is no longer what ships.