The Agent Skills Directory

[SAFE]: The skill implements a local hook script, hooks/quality-gate.py, which uses only Python standard libraries to perform system health checks and enforce project documentation habits.
[DATA_EXFILTRATION]: No network activity was detected. The script does not import networking modules (e.g., socket, requests, urllib) and all output is confined to the local stderr stream for providing feedback to the agent.
[COMMAND_EXECUTION]: The script does not use os.system, subprocess, or other functions to execute shell commands. All operations, such as disk usage and file modification checks, are performed using standard library functions like shutil.disk_usage and os.path.getmtime.
[EXTERNAL_DOWNLOADS]: The skill has no external dependencies and does not download any code or data from the internet.
[REMOTE_CODE_EXECUTION]: No dynamic execution patterns such as eval(), exec(), or runtime compilation were identified.
[INDIRECT_PROMPT_INJECTION]: While the skill processes the agent's transcript, which constitutes an ingestion point for untrusted data, the capabilities are restricted to local metadata checks and regex-based warnings.
Ingestion points: hooks/quality-gate.py reads the session transcript via sys.stdin or a file path provided in a JSON payload.
Boundary markers: None present.
Capability inventory: File read (transcript), metadata read (disk usage, file modification times). No network or write capabilities.
Sanitization: None applied to the transcript content before regex processing; however, the impact is limited to local logging and exit code signaling.

delivery-gate