The Agent Skills Directory

[PROMPT_INJECTION]: The skill captures free-form user responses during the discovery phase and records them in DISCOVERY.json and PROJECT.json, which drive downstream project management. Ingestion points: Free-form answers to interview questions defined in QUESTIONS.md. Boundary markers: No specific delimiters or safety instructions are used when interpolating user-provided data into project state files. Capability inventory: The skill is authorized to write and update project configuration files and delete existing project artifacts (Agenda, Media Specs, etc.) based on the discovery state. Sanitization: No validation or sanitization of user-provided content, such as font names or URLs, is specified before persistence.
[EXTERNAL_DOWNLOADS]: The skill allows for the storage of an External Font Override containing a sourceUrl provided by the user (file: QUESTIONS.md). This introduces a dependency on remote external resources into the project environment.
[COMMAND_EXECUTION]: The agent is instructed to execute a local JavaScript file (scripts/presentation-theme-invalidation.mjs) to calculate file invalidation plans during requirement updates (file: RESTART-GUARD.md). While the script itself is a static part of the skill, it represents a controlled execution of script logic within the agent context.

discover-presentation