The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it processes untrusted source code from the repository to generate its map.
Ingestion points: Local repository source files are scanned using ast-grep (SKILL.md).
Boundary markers: Absent; the skill does not explicitly instruct the agent to ignore instructions embedded within the codebase being parsed.
Capability inventory: The skill executes ast-grep, map-validator, and writes data to local state directories like .claude/repo-map.json (SKILL.md).
Sanitization: Absent; while AST parsing focuses on code structure, embedded instructions in comments or string literals could potentially influence subsequent agent reasoning when the map is referenced.
[COMMAND_EXECUTION]: The skill requires the execution of external CLI tools (ast-grep and map-validator) to function. The instructions explicitly state that these tools should only be run or installed with user consent, which is a security best practice.

repo-mapping