semgrep

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs scanner subagents to git-clone and use public third‑party Semgrep rulesets (e.g., https://github.com/trailofbits/semgrep-rules) as required by references/rulesets.md and references/scanner-task-prompt.md, so the agent fetches and uses untrusted, user‑generated content that directly influences scanning behavior and downstream decisions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly clones and loads third-party Semgrep rulesets at runtime (for example, https://github.com/trailofbits/semgrep-rules), which are fetched and used as required configs that directly control the scanner's behavior/execution, so this is a high-confidence runtime external dependency risk.