The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references the dspy Python library and official documentation located at dspy.ai, which are established resources for language model programming.
[COMMAND_EXECUTION]: Instructions describe how to use DSPy modules like ProgramOfThought and RLM which generate and run Python code. The skill mitigates risks by recommending the use of the sandboxed dspy.PythonInterpreter based on Deno and Pyodide to isolate code execution from the host system.
[PROMPT_INJECTION]: The skill provides patterns for processing large external documents which creates an attack surface for indirect prompt injection.
Ingestion points: Data is ingested into the agent context via the document parameter in RLM calls.
Boundary markers: The provided examples do not use explicit delimiters, relying on the underlying library's prompting logic.
Capability inventory: The agent is given instructions to execute code using the dspy.PythonInterpreter tool.
Sanitization: Security is addressed by explicitly recommending restricted sandboxes that deny filesystem and network access by default.

dspy-reasoning-modules