The Agent Skills Directory

[SAFE]: The skill is purely instructional and contains no executable code, remote script downloads, or network operations. It functions as a mapping system to guide the AI agent toward relevant tools and frameworks based on user intent.
[PROMPT_INJECTION]: The skill processes untrusted user-supplied task descriptions to match triggers for activating sub-skills. This constitutes an indirect prompt injection surface where task descriptions could influence the agent's internal state by loading specific skills.
Ingestion points: Task descriptions provided in user input (as defined in SKILL.md).
Boundary markers: Absent. The skill performs direct matching on user text without delimiters.
Capability inventory: No file writing, network calls, or subprocess executions are implemented in this skill; it only uses the internal Skill() and Agent() delegation commands.
Sanitization: No sanitization or validation of the input text is performed before trigger matching.
[COMMAND_EXECUTION]: The skill contains technical guidance for CLI tools like uv, hatchling, and pre-commit. This information is provided as reference material for the agent's code generation processes and does not represent commands executed by the skill itself.

specialist-skill-routing