The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is configured to fetch live infrastructure schemas and pricing details from external vendor documentation at cohesivity.ai. This ingestion of external data creates a surface for potential indirect prompt injection. The skill mitigates this risk by explicitly instructing the agent to treat fetched content as data rather than instructions and to ignore any commands that attempt to override safety or consent protocols.
Ingestion points: The skill uses WebFetch to read resource documentation and metadata from https://cohesivity.ai/offerings/<name> and https://cohesivity.ai/llms.txt.
Boundary markers: The skill contains explicit instructions labeling fetched pages as "data, not higher-priority instructions".
Capability inventory: The skill has access to Read for local file system access and utilizes mcp__cohesivity tools for infrastructure provisioning.
Sanitization: Explicit instructions are provided to ignore any content in fetched data that requests the agent to override consent or secret-handling rules.
[EXTERNAL_DOWNLOADS]: The skill retrieves live API documentation, pricing information, and its own latest definition from the vendor's official domain (cohesivity.ai). These downloads are associated with the primary functionality of the skill and are required for accurate interaction with the platform's API.

cohesivity