The Agent Skills Directory

Indirect Prompt Injection (HIGH): The skill's core functionality involves fetching untrusted content from external repositories into the agent's context.
Ingestion points: Untrusted source code, comments, and commit messages are ingested via the src search command as described in SKILL.md.
Boundary markers: There are no boundary markers or instructions to the agent to distinguish between data and potential commands within the fetched content.
Capability inventory: The SKILL.md file explicitly mentions integrating results with tools like Bash, Read, and WebFetch, which provides a direct path for injected instructions to trigger file modifications or command execution.
Sanitization: No sanitization or validation of the fetched external content is performed or recommended.
Command Execution (LOW): The skill facilitates the execution of the src CLI tool. While this is the intended functionality, it introduces the standard risks associated with subprocess execution.

sourcegraph-search