The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from the local repository (code files, manifests, READMEs) to generate architecture overviews and author documentation.
Ingestion points: The skill uses rg, semantic_search_nodes_tool, and graph exploration tools to read the contents of the codebase.
Boundary markers: There are no instructions or boundary markers defined to prevent the agent from executing instructions that might be embedded in the codebase files (e.g., hidden within markdown comments or code strings).
Capability inventory: While this skill is primarily read-only for orientation, it is intended to feed collected facts into the AGENTS.md authoring process, which could be manipulated by adversarial content in the source files.
Sanitization: There is no mention of sanitizing, filtering, or escaping the content extracted from the codebase before it is interpreted by the agent.
[NO_CODE]: The skill consists exclusively of markdown instructions and YAML configuration without any bundled scripts or executables, reducing the risk of direct malicious code execution.

explore-codebase