A Minecraft assistant that can interact with the game through MCP (Model Context Protocol) server, providing tools for executing commands, sending messages, managing players, and accessing world information
The server defines 5 tools with moderate structural quality but significant gaps in descriptions, parameter validation, and error handling. Tool names follow verb_noun convention adequately. However, descriptions are inconsistently detailed, parameter schemas lack comprehensive constraints, and no output schemas are documented. The design shows promise in modularity (separate tools for distinct actions) but falls short of production-grade clarity needed for reliable LLM invocation.
Execute a Minecraft command.
Get the tool list in a specific tool extension package.
Get available tool extension packages list. Returns all available tool extension packages and their short descriptions.
Send a message to the game chat.
Teleport a player to specified coordinates.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract return values reliably.
Parameter constraints missing or insufficient. Numeric coordinates (teleport_player: x, y, z) have no min/max bounds; string parameters lack enums or format specifications; message and command lengths unbounded.
No error recovery guidance. Descriptions do not explain what to do if a command fails, a player is not found, or coordinates are invalid. No hint on retryability or user-fixable vs fatal errors.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-21 | D | 51 | <=2025-11-25 | v2 |
| 2026-03-09 | D | 50 | - | v1 |
Tool descriptions lack context for tool selection. Generic phrases like 'Execute a Minecraft command' do not explain when to use this vs alternatives, prerequisites, or side effects. LLM cannot disambiguate from name alone.
Destructive/write operations (execute_command, send_message, teleport_player) marked as WRITE risk but lack dry-run or confirmation step. Agents may issue irreversible commands without safeguards.
Parameter descriptions are inconsistent in detail. Some (e.g., 'client_id', 'target') are terse; others lack actionable guidance on valid values or format.