Agent Enhancer Utilities v1.5.0 packages the project as a reusable reliability sidecar for existing agents.
Highlights:
- Seven inspectable Agent Skills in the canonical
skills/layout. - Versioned manifests for ChatGPT/Codex, Claude Code, GitHub Copilot CLI, Gemini CLI, and generic remote MCP clients.
- Two runnable, standard-library Python examples: a general reliability sidecar and a two-worker checkpoint race.
- Architecture, compatibility evidence, source-tagged discovery, and weekly owned-surface drift checks.
- Updated ChatGPT review bundle with exactly five positive and three negative cases.
Verified before release:
- Repository and deterministic archive validation
- OpenAI plugin validation
- Claude Code 2.1.220 manifest validation
- Isolated GitHub Copilot CLI 1.0.75 install with all seven skills
- Live MCP initialization, six facade tools, 24 free modules, both runnable examples
- GitHub Actions validation on the release commit
All public modules remain free and account-free. Real USDC settlement remains disabled. The hosted implementation is operated separately from this public skills and integration package.