Loading ToolBench
Local-file-first MCP server for creating, reading, editing, importing, and exporting professional XMind mind map files
Server has 14 tools with explicit schemas and descriptions. Naming follows verb_noun pattern (xmind_create, xmind_open, xmind_patch). Descriptions are present (avg ~120 chars) but lack actionable context for LLM selection. Schemas are well-structured with Zod validation, but output schemas are not documented. Parameter descriptions exist but many are generic ('Local .xmind file path'). Error handling is minimal, no recovery guidance or categorization. Missing idempotency hints on write operations and confirmation patterns on destructive tools (xmind_patch, xmind_save_workbook).
Compatibility alias for xmind_create. Generate a professional XMind mind map file from a tree structure or workbook JSON.
Generate a professional XMind mind map file from a tree structure or workbook JSON. Supports themes, markers, relationships, boundaries, summaries, and rich formatting.
Export an XMind file to JSON format. Returns the complete workbook structure as JSON.
Export an XMind file to Markdown format. Converts the mind map structure to a hierarchical markdown outline.
Export an XMind file to OPML (Outline Processor Markup Language) format.
Import a Markdown file and convert it to an XMind mind map. Supports task markers and links.
Import an OPML file and convert it to an XMind mind map.
Output schemas not documented. Tools like xmind_open, xmind_export_json, xmind_patch return complex workbook/result objects, but LLMs have no schema to understand what fields to expect. This forces agents to guess at response structure and breaks downstream tool chaining.
No error recovery guidance. Tools fail silently or with generic errors (e.g., file not found, invalid JSON). Descriptions do not explain what to do on failure, should the agent retry, check the file path, or call a different tool? This violates recovery-guide pattern.
Destructive operations lack confirmation or dry-run. xmind_patch and xmind_save_workbook modify files irreversibly. No dry-run mode documented, no confirmation step, no undo capability. Agents can corrupt files without safeguards.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 59 | 2025-06-18+ | v2 |
Inspect an XMind file for structure, validation, and feature support. Returns detailed analysis of sheets, topics, markers, relationships, and any issues.
List all available XMind markers (priority, task, emoji, flags, stars, etc.) organized by category.
List all available XMind layout structures (mind map, logic chart, org chart, fishbone, timeline, etc.) with descriptions.
List all available XMind themes with their color schemes and properties.
Open and read an existing XMind file. Returns the full workbook structure with all sheets, topics, relationships, and metadata.
Apply a series of patch operations to an existing XMind file. Supports adding, updating, deleting, and moving topics; managing relationships, markers, labels, notes, links, tasks, boundaries, and summaries.
Save a workbook JSON object to an XMind file. Use this after modifying a workbook returned by xmind_open.
Parameter descriptions are generic and lack constraints. 'Local .xmind file path' appears 6 times but does not specify: must file exist? What if path is relative vs absolute? What character restrictions apply? LLMs cannot infer these rules.
Tool descriptions lack actionable context. 'Export an XMind file to JSON format' does not explain WHEN to use this vs xmind_open (which also returns JSON). No guidance on which export format suits which use case. LLMs cannot disambiguate.