Role proof packet · checked May 19, 2026

AgentProof shows the verification layer after agents write code.

I built AgentProof as a public harness for AI-built apps and pull requests: local commands, browser routes, screenshots, console errors, accessibility basics, broken links, launch-readiness score, machine JSON, and handoff notes.

Proof

What to inspect

Artifact CLI + Pages + GitHub Action

Release AgentProof v0.1.7 public repo

Fit Agents, Codex, evals, developer tools

Fit

Current target lanes

AI Deployment Engineer for Codex: proof that real repos can be scoped, checked, and handed off safely.
AI Systems Engineer, Codex Agents: traceability, tool-use failure surfaces, and verified task completion.
Applied AI Engineer, Codex Core Agent: eval thinking around generated code, routes, reports, and user-facing proof.
Developer Productivity / Quality & Developer Tools: CI action, local harness, score, report artifacts, and public docs.

Evidence

Links

OpenAI Codex proof brief Codex deployment memo Cleanroom Intelligence Works repo Public repository Hosted demo report Report reader Pull-request proof demo GitHub Action proof

Commands

Local check

npm install
npm run build
npm test
node dist/src/cli.js audit . --no-browser

Cleanroom

Boundary

Cleanroom Intelligence Works is the legally distinct proof-lab concept around this work. I act as founding CEO for that pre-incorporation build: ship small proof tools, reject fake scale, and keep claims tied to tests, reports, source links, and current role fit.

This page uses OpenAI and Codex only to name public target roles and source context. It is independent from OpenAI and does not use OpenAI logos, product UI, private systems, or affiliation language.

Sources checked: OpenAI Brand Guidelines, OpenAI engineering roles, and OpenAI Emerging Talent. The role-specific follow-up is the Codex deployment memo.