The Agent Skills Directory

[COMMAND_EXECUTION]: The orchestrator script (scripts/doctor.sh) dynamically executes a series of diagnostic bash scripts bundled within the scripts/checks/ directory. These scripts perform operations such as checking Git worktree status and validating JSON configuration structure using jq to ensure repository health.
[INDIRECT_PROMPT_INJECTION]: The skill processes local repository configuration files (.woostack/config.json, .woostack/.gitignore) and project definitions, representing a potential attack surface where malicious data in those files could be included in the diagnostic reports provided to the agent.
Ingestion points: Local configuration and manifest files including .woostack/config.json, .woostack/.gitignore, and workspace manifest files.
Boundary markers: The skill relies on standard diagnostic reporting formats and enforces a human-in-the-loop requirement for all repairs, but does not use cryptographic delimiters for ingested content.
Capability inventory: The tool is limited to local filesystem reads, JSON structure validation, and gated filesystem repairs (such as updating .gitignore or config.json via templates).
Sanitization: Input data is handled using jq to ensure JSON integrity, though specific diagnostic messages derived from file contents are passed through to the agent's output.

woostack-doctor