The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to utilize the uvx disambiguate <term> command to verify vocabulary consistency against project documentation.
[EXTERNAL_DOWNLOADS]: The use of uvx facilitates the downloading and execution of the unversioned 'disambiguate' package from public repositories at runtime.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process potentially untrusted data from project specifications, external URLs, and conversation logs. Ingestion points: Specification file paths, issue numbers, and external URLs provided in context. Boundary markers: The instructions lack explicit delimitation or isolation protocols for untrusted content during processing. Capability inventory: The skill possesses capabilities for reading external file and URL content and publishing processed tickets to project trackers. Sanitization: The skill relies on human-in-the-loop validation, including a 'grilling gate' and 'user quiz' to review generated tickets before publication.

to-tickets