Launching today
Sutura
Verified self-healing CI that proves the fix
8 followers
Verified self-healing CI that proves the fix
8 followers
AI agents make CI pass, but a green check does not prove the failure was fixed. Sutura is a GitHub Action and CLI that reproduces the failure in an isolated sandbox, separates flakes from real failures, searches bounded repairs, rejects green-wash (deleted tests, weakened assertions, relaxed config), then puts the winner through an adversarial audit by NVIDIA Nemotron with GPT-6 Astra and TypeSafe Jev as veto-only second opinions. It opens an evidence-backed PR for a human. Never auto-merges.
Products used by Sutura
Explore the tech stack and tools that power Sutura. See what products Sutura uses for development, design, marketing, analytics, and more.
Engineering & Development 4
Engineering & Development 4
GitHub ActionsAutomate your workflow from idea to production
5.0 (28 reviews)
Sutura ships as a GitHub Action, so Actions is both the runtime and the thing it repairs. It reads the exact failed workflow run, reproduces it, and opens the evidence-backed pull request through the same API. The whole release benchmark and the public demo run on Actions too.
Codex 3.0 by OpenAICodex can now build, test & debug on autopilot
5.0 (95 reviews)
Astra was the coding model this project started with, and it now runs inside the product too: GPT-6 Astra is the second-opinion auditor that re-asks whether a repair fixed the diagnosed failure and can veto the primary verdict. Built with Astra, and running Astra.
Claude CodeAnthropic’s deep-context AI coder
5.0 (694 reviews)
Most of Sutura was written in Claude Code sessions: the RPI workflow, the verification gates, the release tooling, and the Case Lab. The subagent and worktree model let one person ship a verified CI repair pipeline with a real benchmark in weeks.
VercelThe frontend cloud. Creators of Next.js.
5.0 (1K reviews)
The public Case Lab runs on Vercel: static case pages plus the small serverless dispatcher that starts live repairs and enforces the hourly and daily caps. Deploys are one CLI command, and the production environment is what gates live runs to one exact release.
AI Agents 1
AI Agents 1
ElevenLabsCreate natural AI voices instantly in any language
4.9 (205 reviews)
The launch video narration is an ElevenLabs voice reading a 75-second script. It took one pass to get a calm, technical read with the right pauses at the cuts, which is what let a solo maker ship a narrated demo the same evening the video was recorded.