Configuration
LINZA is configured through environment variables in your MCP client. There is no separate YAML config: point it to a folder, choose an embedding provider, and optionally choose the tool set.
Core Variables
| Variable | Meaning |
|---|---|
LINZA_VAULT | Absolute path to the working folder or Markdown vault. Required. |
LINZA_EMBED_PROVIDER | Embedding provider: lmstudio, ollama, or openai. |
LINZA_EMBED_URL | Local or remote embedding endpoint. |
LINZA_EMBED_MODEL | Embedding model name. |
LINZA_EMBED_KEY | API key, if the provider needs one. |
LINZA_TOOL_SURFACE | Tool set: default for normal use, advanced for debugging. |
Tool Set
For normal work, keep LINZA_TOOL_SURFACE=default. The agent sees a compact tool set: diagnostics, index, search, map, review cards, teaching, growth, relationship explanation, and context export.
advanced is for development and diagnostics. It exposes lower-level actions for inspecting local database state, and is usually unnecessary for user workflows.
MCP Configuration Example
{
"mcpServers": {
"linza": {
"command": "linza-mcp",
"env": {
"LINZA_VAULT": "/absolute/path/to/your-folder",
"LINZA_EMBED_PROVIDER": "lmstudio",
"LINZA_EMBED_URL": "http://127.0.0.1:1234/v1",
"LINZA_EMBED_MODEL": "your-embedding-model-name",
"LINZA_TOOL_SURFACE": "default"
}
}
}
}For VS Code / Copilot MCP the shape is almost the same, but the key is servers:
{
"servers": {
"linza": {
"type": "stdio",
"command": "linza-mcp",
"env": {
"LINZA_VAULT": "/absolute/path/to/your-folder",
"LINZA_EMBED_PROVIDER": "lmstudio",
"LINZA_EMBED_URL": "http://127.0.0.1:1234/v1",
"LINZA_EMBED_MODEL": "your-embedding-model-name"
}
}
}
}Embeddings
For local work, LM Studio is usually the simplest path: data stays on your machine, while LINZA gets vectors for search, topic maps, and review cards.
If you change provider, model, or embedding dimension, run a full reindex. Vectors from different spaces cannot be compared reliably.