The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were detected in the skill instructions or reference files. The documentation prioritizes data privacy, instructing the agent to store customer sensitive data in private workspaces rather than public repositories. It also explicitly forbids hardcoding of credentials and requires dynamic discovery of tool schemas. Control flow guides the agent to verify authorizations before making broader API calls, and emphasizes that document instructions must not be treated as authoritative, which provides a layer of defense against indirect prompt injection.

recoup-release-add-release