@yaima3
Yaima ValdiviaFounder▲6
Founder, AgentTrust ID
3d ago · New project · Agents · San Francisco

Shipped: SDLC Pipeline

SDLC Pipeline is a control plane for AI-assisted software delivery. Agents do the work; the pipeline decides what a change must prove before a person merges it, and CI checks the proof. It carries requirement IDs from ticket to merge, stops agents from adding scope silently, keeps tests out of the implementer’s reach, reviews in a fresh context, binds review and QA evidence to the exact tree, and leaves commit and merge to a human. It runs with Claude Code, Codex CLI, GitHub Copilot, Gemini CLI, or any chat tool with repo access.

SDLC Pipeline logoSDLC Pipeline@yaima/sdlc-pipeline · github.com/Yaima/sdlc-pipeline

Comments2

  • @ryanallis
    Ryan Allis
    Founder of AI Lounge and SaasRise, helping founders reach $100M+ exits
    2d ago

    Keeping the final merge in a person's hands, and stopping agents from quietly adding work nobody asked for, hits the two worries engineering leads raise most when they let AI write their code. That kind of control is often what gets a cautious team to say yes to agents in the first place.

  • @yaima
    Yaima Valdivia
    Principal Software Engineer | Founder, AgentTrust ID | AI Agent & LLM Security | Instructional Associate at Georgia Tech
    2d ago

    Those two were the first rules in the pipeline and they are the two that have not changed since. Everything else (review rounds, tiering, the evidence checks) came later from running features and reading the research.

Log in to join the thread.