github-code-review

Fail

Audited by Gen Agent Trust Hub on Oct 6, 2026

Risk Level: HIGHDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The webhook-handler.js example provided in the documentation uses execSync to execute shell commands constructed using unvalidated input from GitHub webhook payloads (event.comment.body). This implementation introduces a command injection vulnerability, potentially allowing any user who can comment on a pull request to execute arbitrary code on the runner.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from GitHub, including PR titles, bodies, and diffs, which are then used to influence the behavior of the AI agents and subsequent repository actions.
  • Ingestion points: PR data is fetched via gh pr view and gh pr diff commands throughout SKILL.md.
  • Boundary markers: None; external PR content is passed directly to the swarm analysis tools as string arguments.
  • Capability inventory: The skill has significant capabilities including gh pr review (commenting, approving, requesting changes), gh pr edit (labeling), and gh pr merge (merging code).
  • Sanitization: The skill lacks explicit sanitization or escaping mechanisms for the PR content it processes.
  • [COMMAND_EXECUTION]: The skill heavily relies on executing shell commands via the GitHub CLI (gh) and npx to perform repository management tasks, including automated code merging and labeling.
  • [REMOTE_CODE_EXECUTION]: The skill frequently uses npx to download and execute the ruv-swarm package. While this tool is associated with the skill's vendor, the use of npx involves fetching and running remote code during execution.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Oct 6, 2026, 12:27 AM