planner

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

• Third-party content exposure detected (high risk: 0.95). The skill reads the outsider-authored GitHub issue body via gh issue view {issue-number} --json ... ,body and then injects the extracted ## Idea and ## Research Summary text into the agents’ LLM prompts (e.g., IDEA: {idea-text} / RESEARCH SUMMARY: {research-summary-text}), which is free-form text authored by someone other than the operating user.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill invokes the GitHub CLI at runtime (gh issue list / gh issue view), fetching issue bodies from the GitHub Issues API (e.g. https://api.github.com), and explicitly injects the fetched sections (## Idea, ## Research Summary) into model prompts, so remote issue content directly controls agent instructions.