mcp-code-execution
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes large external datasets (exceeding 10k rows or 50KB) using automated tool chains and subagents, which creates a potential surface for indirect prompt injection. * Ingestion points: Input data processed by the mcp-code-execution hub and the mcp-patterns module. * Boundary markers: The instructions do not specify the use of delimiters or explicit warnings to the model to ignore embedded instructions in external data. * Capability inventory: The skill utilizes subagent spawning via the Task tool and sequential tool chaining. * Sanitization: No sanitization, escaping, or validation of external content is described in the skill modules.
- [COMMAND_EXECUTION]: The module mcp-patterns.md contains documentation and examples for executing a local script named tools/extracted_tool.py using the Python interpreter.
Audit Metadata