Loading ToolBench
RADKit MCP server container runtime for agentic workflows with command filtering and access control
Two tools with minimal documentation. execute_cli_commands has a basic description but lacks parameter type definitions (input schema shows 'any' and 'sequence' without proper JSON Schema typing). get_access_mode has no parameters but also lacks output schema documentation. Neither tool has descriptions exceeding 100 characters. No error handling guidance, no parameter constraints (e.g., enums for access modes), and no output schema documentation. The profile.py wrapper adds command filtering logic but this is not exposed in the tool definitions themselves.
Run one or more CLI commands against one or more devices with command filtering based on access mode (show, conf, debug, or full)
Build a readable description of the current command access mode
Input schemas lack proper JSON Schema typing. 'devices' parameter typed as 'any' and 'commands' as 'sequence' without element type, constraints, or descriptions.
No output schema documented for either tool. LLMs cannot plan downstream calls or extract required fields. execute_cli_commands returns 'Any' (untyped); get_access_mode returns 'str' with no structure.
execute_cli_commands description (95 chars) lacks error guidance. No mention of what happens when commands are blocked, what exceptions are raised, or how to recover.
No parameter descriptions. 'devices' and 'commands' parameters lack guidance on format, constraints, or examples.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 50 | 2026-07-28+ | v2 |
Access mode filtering (show/conf/debug/full) is enforced in profile.py but not exposed as tool constraints or parameter enums. LLM has no visibility into allowed command prefixes without calling get_access_mode first.