The Agent Skills Directory

[SAFE]: The skill serves as a documentation repository for the 'deepagents' framework, providing patterns for memory management, subagent spawning, and middleware integration.
[SAFE]: Documentation examples for API keys and environment variables use safe placeholders such as 'your-api-key' and 'your-tavily-api-key' rather than hardcoded secrets.
[SAFE]: The framework documentation explicitly recommends security best practices, including human-in-the-loop (HITL) approval gates for sensitive operations like 'delete_file' or 'send_email'.
[SAFE]: The filesystem tools described in the skill include sandboxing capabilities via 'virtual_mode', which normalizes paths and prevents unsafe directory traversal.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of agents that ingest external data (e.g., via 'internet_search' or 'read_file'). While this introduces an attack surface, the documentation includes mitigations such as human approval gates and sandboxed backends. Given this is the primary purpose of the framework, the risk is managed.

deepagents