The Agent Skills Directory

[COMMAND_EXECUTION]: The skill demonstrates how to use the dspy.CodeAct module, which dynamically executes AI-generated Python code. Security is addressed through the use of a sandboxed Deno-based interpreter for the generated code, preventing direct access to the host's filesystem, network, or environment.
[PROMPT_INJECTION]: As a framework for agents that process external data (e.g., via the provided read_file example), the skill describes an architecture with a surface for indirect prompt injection. The documentation provides mitigation strategies, such as setting max_iters, validating tool inputs, and minimizing the capabilities of exposed functions.

dspy-codeact