The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on executing local shell scripts (skill-probe.sh, skill-dashboard.sh, skills-refiner-doctor.sh) to perform its primary functions of discovery, logging, and reporting. These scripts use standard system utilities like stat, chmod, awk, and sha256sum for file management and identity verification.
[DYNAMIC_EXECUTION]: The skill-trace.sh script facilitates the injection of a lightweight bash canary block into other SKILL.md files. This block is intended to be executed by the agent upon skill use to record activation events, providing a mechanism for runtime observability that is clearly documented and reversible via a --strip command.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data in the form of local Markdown files and agent configuration metadata (package.json, settings.json).
Ingestion points: Reads local SKILL.md files across multiple discovery surfaces and parses Pi agent configuration files.
Boundary markers: Uses explicit XML-style markers () to identify and safely manage injected tracing content, preventing confusion with original skill instructions.
Capability inventory: The skill is capable of executing shell scripts and performing atomic file-write operations to update skill files on disk.
Sanitization: Implements a custom character-level escaping routine (sr_json_escape) to ensure skill names, paths, and current working directory strings are safely encoded for JSON reporting output.

skill-debug