The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a Python script located at ${CLAUDE_SKILL_DIR}/../../scripts/fitness-config.py to perform initialization, validation, and saving of configuration files. It also executes git commands such as git rev-parse --show-toplevel and git ls-files to determine the repository structure.
[INDIRECT_PROMPT_INJECTION]: The skill performs an automated scan of up to 40 files within the user's project directory (e.g., README, manifests, entry points) to classify the project into an archetype. Maliciously crafted content in these project files could attempt to trick the agent into misclassifying the project or applying incorrect security and reliability weights.
Ingestion points: Project files read during the 'Evidence mode' (Step 1 and Step 2) including READMEs, manifests (package.json, go.mod, etc.), and source code entry points.
Boundary markers: None present to distinguish between file content and instructions.
Capability inventory: The skill has the ability to write a fitness-config.json file to the filesystem via a resolver script and can trigger other review skills.
Sanitization: No sanitization or filtering is performed on the content read from files before it is used for classification logic.

fitness-config-init