deep-dive

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill explicitly instructs the agent to scan project files (including .env) and "use what you find" to ground actionable recommendations and to include commands/config snippets, without any guidance to redact or avoid printing secret values, which can cause the model to output secrets verbatim.

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

• Third-party content exposure detected (high risk: 0.80). The skill's SKILL.md Step 1b and the "Ground in live docs" principle require using the find-docs skill to fetch current documentation from public third-party websites (examples: Supabase, Bun, Vercel docs), meaning the agent will ingest and act on open web content that can materially influence its recommendations and actions.