Zed's agent starts every thread with no memory
One context server fixes that.
Your decisions. Your stack. Where you left off. Add MemoryRouter as a remote context server and Zed can pull it all back.
Two minutes. Permanently.
14 days free
Full access. Connect it, ask Zed to remember, cancel anytime.
See the difference
You: Finish the caching layer we designed
Zed: I don't see a prior design in this thread. What are you caching, and what is the invalidation strategy?
A new thread, and yesterday's design is gone.
You: Check memoryrouter, then finish the caching layer we designed
Zed: Found it: Redis with a 5-minute TTL, invalidated on write via the event bus. Wiring the cache client now.
One tool call, and it is caught up.
Sound familiar?
Re-pasting design decisions into every new thread
Zed's agent losing the thread when you start a new one
Decisions from last week, gone
With MemoryRouter
search_memories and store_memory show up as tools in the Agent Panel
Ask it to check memoryrouter and it recalls the relevant history
Tell it to remember something and it is there next time
14-day free trial. Cancel anytime.
Connect Zed in two minutes
Zed supports remote context servers with OAuth built in. Add the server, authenticate once, done.
{
"context_servers": {
"memoryrouter": {
"url": "https://mcp.memoryrouter.ai/mcp"
}
}
}Or use the UI: Settings > AI > MCP Servers > Add Server > Add Remote Server.
Zed prompts you to sign in with the standard MCP OAuth flow since no Authorization header is set. Once connected, the status dot next to memoryrouter turns green. Save a unique fact in one thread, start a new one, and ask Zed to check memoryrouter for it. That is the moment it clicks.
Sign up free, then add the server above.
How it works
Without MemoryRouter:
You Zed agent Model Response (gone when the thread ends)
With MemoryRouter:
You Zed agent (calls search_memories or store_memory via the context server) Model Response
This is model-directed recall, not an automatic hook. Zed decides when to call the memory tools, the same way it decides when to use any other context server. Your Zed account, model choice, and inference stay exactly where they are. MemoryRouter only stores and retrieves memories.
One config, no key file
Add a URL, not a secret. Zed handles OAuth automatically for remote servers with no Authorization header configured.
Explicit tools, visible status
search_memories, store_memory, and date_search_memories, with a live status indicator confirming the connection is active.
Forwards to External Agents too
Context servers configured in Zed are forwarded over the Agent Client Protocol, so external agents you run through Zed can use MemoryRouter as well.
Same memory everywhere else
One vault. Connect it to Claude, ChatGPT, Cursor, or Claude Code and the memory follows you between tools.
Connect in two minutes. 14-day free trial.
Start free. Then just $20/mo.
14-day free trial. Cancel anytime.
- 200M memory tokens / mo included
- Unlimited projects
- Same memory in ChatGPT, Claude, Claude Code, and Codex
- 300 to 500 ms average recall on a typical vault
- Your keys, your data, your control
No inference markup, ever.
FAQ
Give Zed a memory tonight
14 days free. Two minutes to connect. Tomorrow's first thread already knows everything from today.