The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of local shell commands and scripts to verify the environment and project state. It utilizes git for version control checks, gh (GitHub CLI) for issue and PR management, and standard Unix utilities like lsof, find, and grep. These operations are used to provide 'ground truth' for the session handoff.
[EXTERNAL_DOWNLOADS]: The skill fetches version information and updates from the author's official GitHub repository (Quriov/Quriov-Skills). It also utilizes the npx skills command for global skill management. These resources originate from the vendor's own infrastructure.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data including project configuration files (CLAUDE.md, AGENTS.md, active-tracks.yaml) and session transcripts in JSONL format.
Ingestion points: Reads project documentation and JSONL transcript files during the handoff process.
Boundary markers: The skill includes explicit instructions for the AI to treat input from task issues or comments as untrusted data ('不可信输入') and specifically directs the agent not to execute instructions found within them.
Capability inventory: Uses subprocess calls for git, gh, bash, and python3, and performs file-write operations to create handoff documentation.
Sanitization: Includes warnings and structural requirements (like the '内化复述'/internalized repetition step) to ensure the agent processes information as data rather than instructions.

handoff