independent-review
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local git commands to determine the scope of changes for review.
- Evidence: Uses
git diff,git status, andgit merge-basewith arguments for commits and branches in SKILL.md. - [EXTERNAL_DOWNLOADS]: The skill bundles agent definitions and orchestration commands sourced from the official Anthropics Claude plugins repository.
- Evidence: ATTRIBUTION.md identifies the source as
anthropics/claude-plugins-officialand confirms the files are byte-identical to the upstream repository. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze untrusted code changes, which presents a surface for indirect instructions within diffs.
- Ingestion points: The skill reads data from
git diffand file contents (SKILL.md). - Boundary markers: No explicit delimiters or instructions to ignore embedded prompts are mentioned in the aggregation logic.
- Capability inventory: The skill can execute shell commands for git operations and invoke multiple sub-agents (SKILL.md, ATTRIBUTION.md).
- Sanitization: No specific sanitization or escaping of the diff content is described before analysis.
Audit Metadata