The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a bundled Python script to discover and parse the metadata of other installed agent skills.
Evidence: SKILL.md directs the agent to run scripts/list_skills.py to obtain full skill descriptions for selection purposes.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the repository being analyzed to generate architecture and data flow documentation.
Ingestion points: Repository documents (e.g., README.md, AGENTS.md), project manifests (package.json, pyproject.toml, go.mod), and framework configuration files.
Boundary markers: Not present; the instructions do not require the use of specific delimiters to isolate external codebase content during analysis.
Capability inventory: The skill performs file reads across the repository, writes to docs/CODEBASE_OVERVIEW.md as described in SKILL.md, and executes the local scripts/list_skills.py utility.
Sanitization: Not mentioned; the agent relies on the content of external files to populate the docs/CODEBASE_OVERVIEW.md template.

learn-codebase