The Agent Skills Directory

[SAFE]: The skill follows security best practices by explicitly instructing the agent to exclude the configuration field when reading source information, as this field contains sensitive credentials.\n- [SAFE]: Destructive actions, such as deleting connections or sources, are protected by a mandatory confirmation flow that requires the agent to first display the resource details to the user.\n- [SAFE]: The skill restricts the agent's scope by explicitly forbidding it from attempting to create or modify connection settings, which limits potential misuse.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests metadata from external systems (e.g., connection names from HubSpot or Salesforce), which is a potential surface for indirect prompt injection.\n
Ingestion points: Data retrieved via api_read for connections and sources (SKILL.md).\n
Boundary markers: None explicitly defined for the interpolation of external system names or statuses.\n
Capability inventory: The skill has access to api_write (sync) and api_delete (deletion) tools.\n
Sanitization: No specific instructions are provided for sanitizing or escaping the data fetched from external sources.

connections