A Codex Skill for lightweight acceptance review and bounded fix loops.
-
Updated
Jul 1, 2026
A Codex Skill for lightweight acceptance review and bounded fix loops.
Pi-native workflow baton runner with per-step model switching and isolated step context.
Claude builds in-session (no claude -p); Codex reviews headless. Every claimed result is re-run and diffed (claimed==actual) so hallucinated numbers die at the gate. A self-healing 6-stage build/review loop that runs until a real problem stops it.
Reusable Codex and Claude agent skills for VRM avatar polish: Blender/MMD inspection, SpringBone hair and skirt physics, clipping QA, backups, and evidence-driven review loops.
Fallibilist Plan → Implement → Review → Fix → Loop skill for robust agent workflows
Spec-gated, dual-review-loop implementation workflow for Claude Code
Privacy-first local MCP hub for coordinating Claude, OpenAI, Gemini, Ollama, and optional cloud council seats.
Add a description, image, and links to the review-loop topic page so that developers can more easily learn about it.
To associate your repository with the review-loop topic, visit your repo's landing page and select "manage topics."