run-repo-cleanup
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses Python's
subprocess.runto execute various Git commands across its scripts (audit-state.py,merge-branches.py,sweep-artifacts.py, etc.). This is the intended functionality for repository management, and the commands are constructed using list arguments to prevent shell injection. - [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it reads and processes external data from Git diffs and commit logs. This is documented as a risk factor; however, the skill mitigates this by instructing the agent to perform manual reviews, use subagent reviews for large changes, and avoid blind-resolving conflicts.
- Ingestion points:
scripts/merge-branches.pyandscripts/audit-state.pyingest data fromgit logandgit diffoutputs. - Boundary markers:
SKILL.mdincludes explicit instructions for human-like review of all changes before merging. - Capability inventory: Subprocess calls are limited to the
gitbinary for repository state management and file operations. - Sanitization: The skill relies on manual and subagent review phases to validate content before it is merged into the base branch.
Audit Metadata