The Agent Skills Directory

[SAFE]: The instructional logic in SKILL.md establishes a clear framework for skill selection without attempting to override agent safety protocols or bypass instructions.
[SAFE]: The Python script scripts/route.py performs local relevance scoring using standard libraries and does not involve any network requests, sensitive file access, or dynamic code execution.
[SAFE]: While the skill processes untrusted user input to determine the appropriate skill, it possesses no dangerous capabilities that could be exploited via indirect prompt injection. 1. Ingestion points: User task descriptions and file paths. 2. Boundary markers: Explicit formatting instructions for recommendations. 3. Capability inventory: Execution of a local, logic-only Python script. 4. Sanitization: None.

skill-router