pr-management
Pass
Audited by Gen Agent Trust Hub on Apr 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to perform Git operations, such as
git status --porcelain, file staging, and resolving merge conflicts within the repository.\n- [REMOTE_CODE_EXECUTION]: The validation workflow includes instructions to "install dependencies if manifests changed" and to "run the narrowest credible lint, typecheck, build, and test sequence." This executes code from the pull request being reviewed, representing a controlled dynamic execution risk associated with the skill's primary purpose.\n- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted PR descriptions and code diffs to make triage decisions. It lacks explicit instructions to sanitize these inputs or boundary markers to isolate potentially malicious commands embedded in the PR data.\n - Ingestion points: PR descriptions and code diffs processed in the "Triage lane for active PRs" section of SKILL.md.\n
- Boundary markers: Absent. No specific delimiters or "ignore embedded instructions" warnings are provided.\n
- Capability inventory: Git commands, file modification, and execution of local build/test scripts.\n
- Sanitization: Absent.
Audit Metadata