The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill identifies and ingests untrusted data from various repository files to ground the generated documentation.
Ingestion points: SKILL.md (Steps 1 and 2) instructs the agent to read README*, manifests, task scripts, design.md, and other repository files to extract context.
Boundary markers: Absent. The instructions do not specify the use of delimiters or provide "ignore embedded instructions" warnings when the agent processes these external files.
Capability inventory: The skill performs file writes to AGENTS.md, CLAUDE.md, CONVENTIONS.md, and .gitignore, and executes shell commands such as git diff --check and openez (if available).
Sanitization: Absent. There is no explicit validation or sanitization of the content extracted from the repository files before it is used to populate documentation, relying entirely on the model's interpretation.

setup-codebase