summarizer

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt mandates extracting and relaying verbatim passages (extract-before-abstract, "What Was Found", full attributions, and relaying agent results), so if a source contains secrets (API keys, tokens, passwords) the LLM would be required to include those secret values in its output, creating an exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and ingests web pages for summarization — e.g., SKILL.md Decision Tree: "Is it a URL? → Load url-summarization skill → Fetch content first, then summarize" and references/fidelity-rules.md requiring WebFetch/read URLs — so it consumes untrusted public web content that could contain injected instructions affecting agent behavior.