The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill utilizes $ARGUMENTS interpolation within a JavaScript block to handle argument parsing. While a standard pattern for skill configuration, this creates a potential injection surface if user-supplied arguments are not sanitized before being substituted into the executable script block.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes untrusted external data, including git history and AST symbols from potentially malicious repositories.
Ingestion points: Git history logs and repository source files (processed for AST symbols).
Boundary markers: No explicit delimiters or instruction-guarding markers are defined in the SKILL.md for the data processing phase.
Capability inventory: The skill reads and writes to local state directories (e.g., .claude/repo-map.json) and interacts with the agent-analyzer tool.
Sanitization: No specific sanitization or validation of the ingested repository content is mentioned.

repo-intel