pr-gardening
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from GitHub Pull Request titles and descriptions.
- Ingestion points: External data enters the context via
gh pr viewcalls inscripts/check-readiness.mjsandscripts/find-candidates.mjs. - Boundary markers: State is tracked using hidden HTML comment markers in Paperclip issue comments.
- Capability inventory: The skill has the ability to create Paperclip issues, post comments, and archive inbox entries via the Paperclip API.
- Sanitization: The
scripts/render-report.mjsfile includes a dedicatedescapeMarkdownTextfunction to escape Markdown control characters in contributor-provided text before it is rendered into reports. - [COMMAND_EXECUTION]: The skill executes shell commands using the
ghCLI to inspect repository state and pull request details. These operations are limited to read-only queries as per thelib.mjsimplementation and instructions. - [EXTERNAL_DOWNLOADS]: The skill performs legitimate network communication with GitHub (via CLI) and the Paperclip API (via
fetch) to perform its intended functions.
Audit Metadata