The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze arbitrary codebase content through tools like Glob, Grep, Read, and the Explore agent. This creates a surface for indirect prompt injection if the target code contains maliciously crafted instructions (e.g., in comments or documentation blocks) intended to influence the agent's behavior.
[INDIRECT_PROMPT_INJECTION]: The skill possesses significant capabilities, including Bash for metadata extraction and Write for document creation, which could be misused if an injection attack is successful.
[INDIRECT_PROMPT_INJECTION]: There are no explicit boundary markers or instructions within the skill to treat codebase content as potentially untrusted data or to ignore embedded instructions found during analysis.
[INDIRECT_PROMPT_INJECTION]: The Explore agent approach increases the depth of ingestion, potentially exposing the agent to more hidden injection points throughout the codebase.

codebase-documenter