pr-autopilot

Pass

Audited by Gen Agent Trust Hub on Jul 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill is designed to autonomously execute build, test, and lint commands defined within the repository's configuration files (e.g., package.json scripts, Makefile targets, workflow files). This presents a risk of arbitrary code execution if a malicious pull request branch modifies these configuration files. The skill includes defensive instructions advising the user to only run it on trusted branches and to avoid untrusted forks.
  • [PROMPT_INJECTION]: The skill has a high surface area for Indirect Prompt Injection as it processes external data from PR comments.
  • Ingestion points: Step 1 fetches all unaddressed reviews and comments from humans and bots via the GitHub API.
  • Boundary markers: The skill explicitly defines a 'Trust boundary' and instructs the agent to treat fetched content strictly as code feedback data, not as new instructions.
  • Capability inventory: The skill can perform file system writes, git commits, git pushes, and execute local repository scripts.
  • Sanitization: The agent is instructed to ignore any directives embedded in fetched content that fall outside of code feedback, such as requests to access secrets or change git remotes.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 24, 2026, 03:05 PM
Security Audit — agent-trust-hub — pr-autopilot