ln-913-community-debater

Pass

Audited by Gen Agent Trust Hub on May 12, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches reference documents and templates from the author's repository (levnikolaevich/claude-code-skills) when local shared files are not available.
  • [COMMAND_EXECUTION]: Utilizes the GitHub CLI (gh) via Bash to execute GraphQL mutations for creating discussions and posting comments on GitHub. All operations are performed using the discoverable repository context.
  • [PROMPT_INJECTION]: The skill processes user-supplied arguments and repository content (via Grep) to compose discussion bodies. While this presents an indirect prompt injection surface, the skill implements a mandatory human-in-the-loop checkpoint requiring explicit user approval before any content is published to GitHub. It also includes a 'Fact-Check' phase to verify the accuracy of the generated content against the local codebase.
Audit Metadata
Risk Level
SAFE
Analyzed
May 12, 2026, 06:28 AM