Client
Memtro for Zed
Add Memtro to Zed's settings as an OpenAI-compatible provider and as a remote MCP server for the agent.
Zed's agent panel takes OpenAI-compatible providers under language_models.openai_compatible and remote MCP servers under context_servers, both in settings.json. Memtro fits both: the provider for a model with your memory and routing behind it, the server for memory and connector tools the agent can call. The key goes in the provider's settings UI or an environment variable, never in settings.json.
What you get
- Provider with api_url https://platform.memtro.com/v1 and models auto and memtro, tools enabled so the agent can call them
- Remote MCP server with a bearer header, no local process
- Agent threads that check the ticket, the CRM or the ledger before editing
How to connect
- 01settings.json → "language_models": { "openai_compatible": { "memtro": { "api_url": "https://platform.memtro.com/v1", "available_models": [{ "name": "auto", "display_name": "Memtro auto", "max_tokens": 128000, "capabilities": { "tools": true } }] } } }
- 02Enter the key under the Memtro provider in the agent panel's settings, or set MEMTRO_API_KEY (Zed names the variable after the provider id)
- 03settings.json → "context_servers": { "memtro": { "url": "https://platform.memtro.com/api/mcp", "headers": { "Authorization": "Bearer sk-memtro-…" } } }
Example queries
- › Search memtro for who owns the billing migration
- › auto:agent as the model name runs a full investigation
Why it is better with memory
Through Zed you get the same memory, connector tools, routing and jobs as every other client, so what one person learns in one tool is available to the whole team in all of them.