The Agent Skills Directory

[SAFE]: The skill consists entirely of instructional markdown and a configuration file. It contains no executable scripts, binaries, or commands that interact with the system or network.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-provided intent to generate structured goal prompts. While it acts as a refinement mechanism, it remains an ingestion point for untrusted data. 1. Ingestion points: User-provided fuzzy intent and interview responses processed in SKILL.md. 2. Boundary markers: Instructions enforce a rigid six-field structure (Outcome, Success condition, Scope boundary, Context, Validation loop, Stop and pause rules) to delimit the goal. 3. Capability inventory: Mentions facilitating tool calls for /goal, create_goal, set_goal, and set_goals. 4. Sanitization: No explicit programmatic sanitization is defined, though the structured output format is intended to reduce ambiguity.

goal-griller