claudex-loop
claudex-loop hardens project plans through a four-phase adversarial review process where two AI models (Claude and OpenAI Codex) interrogate and critique a plan before any code is written, then swap roles during build and cross-inspection to ensure the builder is never the grader. The plugin enforces security controls via read-only sandboxing, role-based access gates, allowlist safety guards, and deterministic code review with acceptance criteria across definition-of-done, quality, security, documentation, and tests.
Install
What it's made of
2 agents · 7 skills · 2 hooks
- Commands
- 0
- Agents
- 2
- Skills
- 7
- MCP servers
- 0
- Hooks
- 2 · PreToolUse
What it needs & plugs into
- API keys
- none
- Paid services
- OpenAI Codex
- External tools
bash- Talks to
- nothing external detected
Analyzed . Facts extracted from the plugin's files. Prose generated by claude-haiku-4-5-20251001.
Is this your plugin?
Claim it to keep the card accurate and enter the weekly contest. Requires signing in as the GitHub owner (ujconsulting).