The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local scripts (registry-events.py) and standard environment discovery commands (git rev-parse) to manage its state and directory structure. These operations are restricted to the local environment and intended for skill functionality.- [INDIRECT_PROMPT_INJECTION]: The skill processes external channel exports and messages. It significantly reduces this risk by instructing the agent to treat this data as 'untrusted evidence' and verifying it against a formal schema and catalog before processing.- [CREDENTIALS_SAFE]: The instructions explicitly state that host-capability values must never enter request JSON, files, or logs, which is a best practice for preventing credential leakage.- [DATA_EXPOSURE]: The skill contains logic to minimize the collection of person data during UGC and advocate registration, aligning with data minimization principles.

channel-registry