git-hunk
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill defines a workflow where the agent executes shell commands (
git hunk <subcommand>) to interact with a git repository. This behavior is the primary intended function of the skill.\n- [INDIRECT_PROMPT_INJECTION]: The skill's core functionality involves reading and displaying content from a git repository, which creates a surface for indirect prompt injection attacks if the repository contains malicious instructions.\n - Ingestion points: The
listanddiffcommands read file content and commit summaries from the local git repository (SKILL.md, references/commands.md).\n - Boundary markers: While the output format is described (columnar and unified diff), there are no specific markers or instructions telling the agent to treat the repository content as data and ignore potential commands embedded within it.\n
- Capability inventory: The agent has the ability to stage changes, create commits, and restore/revert files in the worktree based on its interpretation of the diffs (SKILL.md, references/commands.md).\n
- Sanitization: The documentation does not mention any sanitization or validation of the repository content before it is presented to the agent.
Audit Metadata