research
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use
git log,git blame,gh pr view, andgrepto retrieve information about the project's history, pull requests, and file contents. These tools are used appropriately for research and information retrieval purposes.- [INDIRECT_PROMPT_INJECTION]: The skill creates an ingestion surface by reading untrusted data from external sources such as vendor documentation, specifications, live endpoint responses, and team chat logs. - Ingestion points: SKILL.md (Step 2) and references/why.md (Sources section).
- Boundary markers: None present to distinguish external data from agent instructions.
- Capability inventory: The skill can execute shell commands (
git,gh,grep), read local files, and write summary notes to the filesystem. - Sanitization: No explicit sanitization or escaping of external content is defined, which is a potential surface for indirect injection, though typical for research-focused agents.
Audit Metadata