MiMo Code Troubleshooting

# Option A: Use nvm (recommended)
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.7/install.sh | bash
nvm install 20; nvm use 20
npm install -g @mimo-ai/cli

# Option B: Fix npm global permissions (quick)
mkdir -p ~/.npm-global
npm config set prefix ~/.npm-global
echo 'export PATH=~/.npm-global/bin:$PATH' >> ~/.zshrc
source ~/.zshrc

# Add npm global bin to PATH
echo 'export PATH="$(npm config get prefix)/bin:$PATH"' >> ~/.zshrc
source ~/.zshrc

# Or locate and add manually
npm config get prefix
# then add that path + /bin to your PATH

# Using nvm
nvm install --lts; nvm use --lts

# Using Homebrew (macOS)
brew uninstall node; brew install node

# Using apt (Linux)
curl -fsSL https://deb.nodesource.com/setup_lts.x | sudo -E bash -
sudo apt install -y nodejs

# Or skip Node.js entirely:
curl -fsSL https://mimo.xiaomi.com/install | bash

# List providers and stored credentials
mimo providers list

# For a custom provider, check options.apiKey in .mimocode/mimocode.jsonc
# (or ~/.config/mimocode/mimocode.jsonc): no extra spaces or quotes
# OpenRouter keys start with "sk-or-v1-"
# DeepSeek keys from platform.deepseek.com
# Anthropic keys start with "sk-ant-"

# Start Ollama
ollama serve

# In a new terminal, verify
ollama list

# Pull a model if needed
ollama pull qwen2.5-coder:14b

# Correct baseURL values:
#   OpenRouter:   https://openrouter.ai/api/v1
#   DeepSeek:     https://api.deepseek.com
#   Local Ollama: http://localhost:11434/v1

# Verify the endpoint and key outside MiMo Code:
curl https://openrouter.ai/api/v1/models -H "Authorization: Bearer your-key"

# For OpenRouter: check usage at openrouter.ai/usage
# For DeepSeek: top up at platform.deepseek.com
# For Token Plan: upgrade or wait for reset

# Switch to a cheaper model temporarily: pick one with /models in the TUI,
# or change "model" in mimocode.jsonc (for example "custom/deepseek-chat")

# Or turn off Max Mode to reduce token usage:
# remove the "experimental.maxMode" block from your config

# Check your internet connection
curl -I https://api.deepseek.com

# If behind a proxy, set:
export HTTP_PROXY=http://proxy:port
export HTTPS_PROXY=http://proxy:port

# Some regions block API providers; use a VPN
# Or switch to another provider with /connect, or change "model" in mimocode.jsonc

# Option 1: log in or add an API key from inside MiMo Code
mimo
/connect

# Option 2: define an OpenAI-compatible provider in .mimocode/mimocode.jsonc
# (see the Install Guide, step 4), then check it:
mimo models

# In mimocode.jsonc, "model" is <provider-id>/<model-id> (only the first "/" splits them),
# and the key under provider.<id>.models must be the upstream model ID.

# OpenRouter model IDs (check openrouter.ai/models):
#   deepseek/deepseek-chat
#   anthropic/claude-sonnet-4
#   openai/gpt-4.1
#   qwen/qwen-2.5-coder-32b-instruct

# DeepSeek direct: deepseek-chat
# Local Ollama: qwen2.5-coder:14b

# List the models MiMo Code can see:
mimo models

# 1. Run /init in project root
mimo
/init

# 2. Create or update AGENTS.md
# Add project rules, conventions, tech stack

# 3. Review the project memory the agent keeps
#    (stored under the MiMoCode data directory, ~/.local/share/mimocode/memory/)

# 4. Use /dream to consolidate after long sessions
/dream

# 5. If the context is full, rebuild it from the latest checkpoint
/rebuild

# Max Mode is experimental and off by default. It is switched on by this block in mimocode.jsonc:
# { "experimental": { "maxMode": { "candidates": 5 } } }

# Turn it off: remove the "experimental.maxMode" block from the config.
# Fewer parallel candidates cost fewer tokens (the default is 5).

# For routine tasks like comments or formatting, keep it off

# Disable telemetry
export MIMOCODE_ENABLE_ANALYSIS=false

# Add to shell profile for persistence:
echo 'export MIMOCODE_ENABLE_ANALYSIS=false' >> ~/.zshrc

# For fully private work, use a local model (see the Install Guide, option C)

# List configured MCP servers and their status:
mimo mcp list

# Add a server interactively:
mimo mcp add

# Or declare it under "mcp" in .mimocode/mimocode.jsonc:
# { "mcp": { "filesystem": { "type": "local", "command": ["npx", "-y", "@modelcontextprotocol/server-filesystem", "/path/to/allowed/dir"] } } }

# OAuth-enabled remote server: mimo mcp debug <name>

# Test your MCP server independently:
npx @modelcontextprotocol/server-filesystem /path/to/allowed/dir

# Common MCP server examples:
# Filesystem: npx @modelcontextprotocol/server-filesystem /path
# GitHub: npx @modelcontextprotocol/server-github
# Postgres: npx @modelcontextprotocol/server-postgres postgresql://...