github-issues

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

• Third-party content exposure detected (high risk: 0.85). The skill’s required runtime fallback path uses WebSearch/WebFetch (and firecrawl scraping) to retrieve GitHub issue pages from the public web, ingesting outsider-authored issue titles/bodies/comments as readable text into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's web-fallback performs runtime WebFetch calls that retrieve remote content which is then directly parsed via model prompts (e.g., "WebFetch: https://github.com/anthropics/claude-code/issues/11984" and "WebFetch: https://api.github.com/repos/anthropics/claude-code/issues?state=all&per_page=10"), so external URLs are used at runtime to supply content that controls the agent's prompt/context.