gate

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs agents to paste verbatim PR/Linear/session contents into a context bundle and to emit verbatim code excerpts/evidence in reviewer outputs, so any secrets present in those sources would be ingested and reproduced by the LLM with no redaction—creating a high exfiltration risk.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 1.00). The skill explicitly fetches user-generated content from third-party sources — e.g., GitHub PR body/comments via gh pr view and Linear issue + comments via linear-cli (assembled into /tmp/gate-context-bundle.md by the context-fetcher in Step 3g) — and that bundle is read by the context-checker (Step 6) to influence auto-apply decisions and the gate verdict, creating a clear path for indirect prompt injection.