Loading ToolBench
MCP server providing system information, process management, and file cleanup tools
Server has 6 tools with basic descriptions and parameter schemas, but significant gaps prevent higher scores. Tool names follow verb_noun convention (good), but descriptions are minimal (avg 60 chars, well below 194-char baseline). Parameter descriptions are sparse or missing entirely. No output schemas documented. Destructive tools (delete_file, terminate_process_tool) lack confirmation patterns and error guidance. No tool annotations (readOnlyHint/destructiveHint). Error handling is absent, no recovery guidance, no actionable error messages. Composition is reasonable (single responsibility per tool), but chaining support is weak (no IDs returned in responses visible in code).
Delete a file from the system
Check for processes consuming excessive resources.
Return the first running processes in the system
Scan a directory for duplicate, large, or old files
This tool returns system information like CPU, RAM, Disk usage, etc.
Terminate a process after validation and user confirmation
Destructive tools lack confirmation and error guidance. delete_file and terminate_process_tool accept a 'confirmed' boolean but provide no recovery guidance if confirmation fails or operation errors. Error responses must tell the LLM what to do next.
No output schemas documented. Tools return data (FileInfo, ProcessInfo, system info) but the response structure is not visible in tool definitions. LLMs cannot plan downstream calls or extract required fields without documented schemas.
Parameter descriptions are minimal or missing. 'mode' in search_cleanup_candidates has a description, but many parameters lack context. 'confirmed' parameter in delete_file and terminate_process_tool has no description explaining what happens if false.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 47 | <=2025-11-25 | v2 |
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Destructive tools (delete_file, terminate_process_tool) should be marked with destructiveHint=true so agents know they have irreversible consequences.
No error handling or recovery guidance. Tools return success/failure but do not guide the LLM on what to do next. E.g., if delete_file fails due to permission denied, the response should suggest alternatives or next steps.