dspy-rlm-module
Pass
Audited by Gen Agent Trust Hub on Apr 22, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the dspy.RLM module, which executes Python code in a sandboxed environment. By default, this environment is powered by Deno and Pyodide (WASM), ensuring that the code has no access to the host's filesystem, network, or subprocesses.
- [EXTERNAL_DOWNLOADS]: The documentation directs users to install Deno, a reputable and widely-used runtime for JavaScript and TypeScript.
- [PROMPT_INJECTION]: The module is designed to ingest and process large contexts such as logs or entire codebases. This functionality presents a surface for indirect prompt injection if the input data contains malicious instructions. The skill documentation explicitly addresses this risk by warning users against providing unsafe tools like subprocess.run to the RLM. 1. Ingestion points: Untrusted data enters the agent context through the context variable in the rlm call (observed in SKILL.md and example_rlm.py). 2. Boundary markers: The provided examples do not use explicit delimiters to separate the context data from the system prompt. 3. Capability inventory: The skill provides a Python REPL; the documentation demonstrates how users can add capabilities like read_file or grep via the tools parameter (reference.md). 4. Sanitization: No specific sanitization or filtering logic is provided for the input context.
Audit Metadata