The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash, Read, Grep, and Glob tools to scan and analyze the contents of the ~/.claude/ directory.
Evidence: The skill instructions specify scanning the ~/.claude/ directory and using tools to calculate scores based on the presence and complexity of local configuration files.
[DATA_EXFILTRATION]: While the skill reads local configuration metadata, there are no network operations or external data transmission patterns detected.
Ingestion points: Processes configuration files found in ~/.claude/ (SKILL.md, CLAUDE.md, etc.).
Boundary markers: Absent; the skill processes the files directly to determine scoring.
Capability inventory: Utilizes Bash, Read, Grep, and Glob for file system inspection.
Sanitization: Not explicitly mentioned, but the skill's primary function is diagnostic reporting rather than executing the content of the files it reads.

harness-audit