Rei-Clawd

**Judgement:** Rei-Clawd is a focused, well-scoped skill that reliably wires Rei Qwen3 Coder into the Clawdbot/OpenClaw stack. It’s strong on integration details, usability, and troubleshooting, making it a solid choice whenever you need to use coder.reilabs.org as an OpenAI-compatible backend. **What it does well** - Provides both **scripted and agent-driven setup**, minimizing manual JSON edits. - Documents **manual config** clearly, including `baseUrl`, `apiKey`, `api`, `headers`, model metadata, and agent defaults. - Includes **switching and revert scripts**, which lowers the risk of breaking an existing setup. - Addresses common failure modes: 403s from missing `User-Agent` and “Model not allowed” due to missing allowlist entries. **Risks / Limitations** - Depends on an external service (`coder.reilabs.org`): outages or API changes could break functionality without updates to the skill. - API key handling is assumed; users must still manage keys securely (environment, shell history, logs). - Functionality is narrow: it configures and switches providers; it does not add higher-level workflows or safeguards around the model itself. **Recommended scenarios** - You are setting up **Rei Qwen3 Coder as a primary or secondary coding model** in an existing Clawdbot deployment. - You need an **easy way to toggle** between Rei and another model (e.g., Opus) for side-by-side evaluation. - You’re **debugging 403 or model-allowlist issues** when calling Rei from Clawdbot and want a known-good reference configuration. - You want to let non-expert users configure Rei by simply **asking an agent** instead of editing JSON or writing scripts.