interrogate
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes
git diffoutput which acts as untrusted data. A malicious contributor could embed prompt instructions in the code to manipulate the agent's review and editing behavior.\n- Ingestion points: Data fromgit diffvia$ARGUMENTSinSKILL.md.\n- Boundary markers: None present.\n- Capability inventory: Code editing, execution oftypecheckand test commands.\n- Sanitization: None present.\n- [COMMAND_EXECUTION]: The skill directs the agent to execute shell commands liketypecheck,tests, and a utility scriptscripts/diff-stats.sh. These are typical for development environments but represent a command execution surface.
Audit Metadata