The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface where instructions embedded in external HubSpot workflow metadata could potentially influence agent behavior during the audit.
Ingestion points: The skill fetches workflow metadata (names, types, and descriptions) from the HubSpot v4 Automation API in SKILL.md (Stage 2).
Boundary markers: None present; the agent is not explicitly instructed to ignore potential instructions embedded within the workflow names it processes.
Capability inventory: The skill includes scripts using requests to perform GET, PUT, and DELETE operations on HubSpot resources.
Sanitization: No programmatic sanitization is applied to the retrieved metadata, though the workflow includes a manual confirmation step before execution.
[SAFE]: The skill manages credentials securely by instructing the user to store them in a .env file rather than hardcoding them.
[SAFE]: All network operations are directed at HubSpot's official API domain (api.hubapi.com), which is a well-known service.

cleanup-workflows