The Agent Skills Directory

[SAFE]: The skill provides instructions and a setup script for the deepagents framework, which is associated with the trusted LangChain ecosystem. No malicious code or obfuscation techniques were detected in the skill body or scripts.
[SAFE]: The setup script (scripts/setup.sh) uses standard package managers (pip, uv) to install the framework and its dependencies without performing any suspicious network operations or remote script execution.
[SAFE]: The documentation explicitly highlights security risks associated with the LocalShellBackend and provides clear examples of how to implement Human-in-the-loop (HITL) approval via interrupt_on to mitigate unauthorized command execution.
[INDIRECT_PROMPT_INJECTION]: As a framework for building file-aware agents, it inherently possesses an attack surface where processed files could contain malicious instructions. The skill mitigates this by providing built-in mechanisms for human approval before sensitive tool calls are executed.

deepagents