fix-ci

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The prompt explicitly instructs producing CI summaries that include "relevant env vars" and the "full text" of error messages (with no paraphrasing), which can contain API keys, tokens, or other secrets and thus would require the LLM to output secret values verbatim without guidance to redact them.