The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified. The skill is authored by the platform vendor and all remote resources are hosted on official or well-known trusted domains.- [REMOTE_CODE_EXECUTION]: The skill provides instructions for installing official platform tools. Specifically, it references installing MCP servers from https://api.baseten.co/mcp and provides an example for installing Ollama via a remote script at https://ollama.com/install.sh. These are standard, documented installation methods for the respective services and do not pose a threat in this context.- [PROMPT_INJECTION]: The skill's evaluation suite (evals/evals.json) contains a preamble intended to override standard tool-use confirmation behavior during automated tests. It instructs the agent to invoke tools without asking for permission. This is a common pattern in testing environments to ensure automated flows can complete without human intervention and does not affect the safety of the skill during normal use.- [DATA_EXPOSURE]: The skill demonstrates safe practices for secret management. It instructs users to use environment variables for API keys and explicitly warns against hardcoding secrets in configuration files.

baseten