Hey PH 👋 Solo founder here (Honduran engineer in Texas, 25+ years building for publicly traded companies + US defense supply chains). I built this because of one question nobody could answer: when an AI agent writes half your code, who approved each change, and can you prove it? The three big labs all shipped AI code scanners this year. They FIND problems. None of them will refuse a merge, and none records who said yes. linebreak-gate is the enforcement half: a fail-closed gate in CI (if the scanner crashes, it fails closed, never shrugs) plus acceptance criteria a named human signs. Your agents read the spec over MCP and can check their own work, but structurally cannot edit what a human approved. The gate is free forever, git-only, no account: uv tool install linebreak-gate, ~10 minutes to your first gated PR. How we eat: signed, offline-verifiable approvals for teams that answer to auditors ($99/mo per team). Everything ships through its own gate (dogfooding is the whole point). I'll be here all day: ask me anything, especially the hard stuff.

LineBreak Gate

The CI gate that says no to unapproved AI code

The CI gate that says no to unapproved AI code