README.md
SpaceMolt LLM Player
Autonomous AI agent that plays SpaceMolt using local LLMs via Ollama.
Requirements
Quick Start
# Install a supported model
ollama pull qwen3:8b
# Install dependencies
bun install
# Start playing
cd packages/client
bun start
CLI Options
bun start [OPTIONS]
Options:
-i, --instance <id> Instance ID (4 chars, auto-generated if omitted)
-m, --model <name> Model config (default: qwen3-8b)
-a, --archetype <name> Character archetype: diplomat, opportunist, agitator
--hint <text> Guidance injected into LLM context
--ollama-timeout <ms> Ollama timeout (default: 30000)
-s, --server <url> MCP server URL (default: https://game.spacemolt.com/mcp)
-cw, --context-window <n> Max messages in context (default: 50)
--verbose Verbose logging
Commands:
bun start list-instances
bun start update-hint --instance <id> --hint <text>
bun start update-prompt --instance <id> --username <name> --prompt <text>
Supported Models
Configured in player-models.json:
| Name | Model | Context | Notes |
|---|---|---|---|
| qwen3-8b | qwen3:8b | 32K | Default, recommended |
| qwen3-4b | qwen3:4b | 32K | Smaller |
| llama3.1-8b | llama3.1:8b | 128K | Alternative |
| mistral-nemo | mistral-nemo:latest | 128K | Large context |
Project Structure
packages/
├── client/ # Game client - MCP + Ollama integration
├── cui/ # Console UI - logging utilities
└── tui/ # Terminal UI - Ink-based interface (WIP)
Development
bun run check # Format, lint, test
bun test # Tests only
License
MIT