The Agent Skills Directory

[Indirect Prompt Injection] (HIGH): The skill is intended to read and process source code to generate documentation, creating a significant attack surface for instructions embedded in data.
Ingestion points: Uses Read, Grep, and Glob tools to ingest untrusted source code and comments.
Boundary markers: No instructions are provided to the agent to distinguish between code to be documented and potential instructions embedded within that code.
Capability inventory: Possesses Bash, Write, and Edit tools, allowing for arbitrary command execution and file system modification if an injection is successful.
Sanitization: No sanitization or validation of the ingested content is performed before processing.
[Command Execution] (MEDIUM): The inclusion of the Bash tool is excessive for a documentation skill. This significantly escalates the potential impact of any prompt injection attack, as the agent can be coerced into executing shell commands in the host environment.

tasks-documentation