LoopX v0.1.0 — first public release
LoopX v0.1.0 Release Notes
LoopX v0.1.0 is the first public release of the cross-tool quality gate skill package for Codex and Claude Code.
Highlights
- Shared
loopx/skill source for Codex and Claude Code. - Stage-based workflow from requirement intake through release readiness.
- Local controller for persistent run state under
docs/loopx/runs/<run_id>/. - Strict validation (
validate --strict,gate) with human confirmation gates for interview and solution review. - Risk-based mode selection (LIGHT / STANDARD / FULL); LIGHT mode supports legal SKIPPED review gates via
MODE_SKIPPABLE_STAGES. - Repair loop with per-stage auto-repair limit: the 3rd failed auto repair records
BLOCKED. - JSON schemas for state, stage results, tracking, mode selection, specs, and interviews.
- Role instructions for requirement, design, review, testing, implementation, validation, and release stages.
- Templates for staged artifacts and final reports.
- 71 regression tests for the controller and package structure.
Install
git clone git@github.com:rye567/loopx.git
cd loopxLink or copy loopx/ into your tool skills directory as loopx.
Suggested Announcement
I built LoopX: quality gates for AI coding agents.
It is a Codex / Claude Code skill package that forces risky AI coding work through requirement interview, spec, review gates, test design, implementation, validation, and release readiness before claiming done.
Useful when agents touch API contracts, permissions, tenant boundaries, state machines, SQL/MQ flows, or cross-module changes.
Repo: https://github.com/rye567/loopx
Suggested GitHub Topics
codex, claude-code, ai-agents, agent-workflow, quality-gate, developer-tools, automation, python, prompt-engineering