land-it
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local CLI tools to automate pull request management and repository operations.\n
- Evidence: The script
scripts/land-gate.mjsusesexecFileSyncto rungh(GitHub CLI) for PR inspection and merging, andgitfor tree comparisons and fetching.\n- [INDIRECT_PROMPT_INJECTION]: The skill incorporates untrusted data from external sources that can influence the agent's actions or the commands it executes.\n - Ingestion points: Pull request bodies (parsed by
scripts/land-gate.mjs), repository README files (instructed to be read inSKILL.mdStep 3), and conversation history.\n - Boundary markers: Uses specific markdown headers like
**Review.**and**Merge danger.**as delimiters in PR descriptions to identify control metadata.\n - Capability inventory: The skill can merge pull requests, update branches, and run repository-defined "post-merge steps" or "acceptance checks".\n
- Sanitization: The
land-gate.mjsscript validates the PR state using GitHub API metadata (e.g.,mergeable,reviewDecision) in addition to parsing text-based receipts, which provides a layer of verification against simple text manipulation.
Audit Metadata