Loading ToolBench
mvmt provides 5 file-system tools with clear verb-noun naming (search, list, read, write, remove). All tools have descriptions (avg 95 chars) and input schemas with typed parameters. However, schemas lack depth: parameter descriptions are minimal (10-30 chars), output schemas are undocumented, and error handling is absent. The 'search' tool accepts optional 'mounts' array but provides no guidance on valid mount names. 'write' includes an 'expectedHash' parameter for conflict detection (good pattern), but no documentation of the hash algorithm or failure modes. No tool describes what it returns or how to chain results. Descriptions meet the 10-1024 char baseline but fall short of LLM-optimized 50-200 char range.
List files and directories in mounted folders or at the root to discover available mounts
Read the full content of a text file from a mounted folder
Delete a file from a writable mounted folder
Search indexed text files across mounted folders using keyword matching
Write or overwrite a text file in a writable mounted folder
Output schemas completely undocumented. No tool describes what fields it returns, forcing LLMs to guess structure and preventing downstream tool chaining.
Parameter descriptions are minimal (10-30 chars). 'mounts' param in search tool lacks guidance on valid mount names or discovery method. 'limit' lacks min/max bounds.
No error handling guidance. Tools provide no recovery hints (e.g., 'path not found, call list() to discover available paths'). LLMs cannot self-correct on failures.
'write' tool's 'expectedHash' parameter lacks documentation of algorithm (SHA256 assumed but not stated), failure behavior, or when to use it.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 59 | 2026-07-28+ | v2 |
No pagination or result limits documented for 'search' and 'list' tools. 'search' mentions 'default 8, max 20' in description but not in schema constraints.