MCP server that exposes Minecraft game state and control via HTTP. Provides tools to execute code in the running game, run Minecraft commands, and capture screenshots.
Three tools with minimal documentation. All have descriptions (10-50 chars), but lack parameter-level descriptions entirely. No input schemas visible in source, only parameter names and types inferred from brief tool descriptions. execute_code and run_command are high-risk irreversible/write operations with no error handling guidance, confirmation patterns, or recovery hints. take_screenshot is read-only but still lacks schema detail. No output schemas documented. Tool names are verb-based (execute, run, take) but descriptions are too terse to guide LLM selection. No pagination, no structured output examples, no field documentation.
Runs Kotlin code inside the running game via ReplBridge
Runs a Minecraft command on the server
Captures the client's framebuffer as an image
No input schemas visible in source code. Tool definitions show only parameter names ('code', 'command') without JSON Schema type/description/constraint metadata. Cannot verify schema completeness.
execute_code and run_command are irreversible/write operations (marked IRREVERSIBLE and WRITE risk) with no error handling, confirmation pattern, or recovery guidance. LLM has no way to know consequences or how to recover from mistakes.
Parameter descriptions missing entirely. 'code' parameter in execute_code lacks guidance on Kotlin syntax, available APIs, or constraints. 'command' in run_command lacks Minecraft command format hints. LLM must guess valid input.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 38 | <=2025-11-25 | v2 |
No output schemas documented. LLM cannot plan downstream operations or extract structured data from tool responses. take_screenshot returns an image but format/encoding is undocumented.
Tool descriptions are 10-50 characters, below the 34-char p10 baseline and well below the 194-char average for production tools. Descriptions lack WHAT/WHEN/WHY context needed for LLM selection.