Gnosem

Prototype · alpha

One memory. Every model.

Cross-vendor AI memory over the Model Context Protocol.

One persistent memory store that Claude, ChatGPT, Cursor, Windsurf, Kimi, Gemini — anything that speaks MCP or can call an HTTP tool — can read and write to. Write a fact in one, recall it from any other.

See pricing →Read the launch story →Sign in →

---

Try semantic search

Search a demo memory graph — no signup, no config. This is the same engine you'd wire your models into.

Search →

Searching a demo memory graph — no signup

See what's in it

Loading demo memories…

1. Sign up

curl -sX POST https://gnosem.dev/signup
-H "Content-Type: application/json"
-d '{"email":"you@example.com"}'

Returns a one-time api_key. Save it — it's not shown again.

2. Install

npx gnosem-install

Auto-detects and configures Claude Desktop, Claude Code, Cursor, Windsurf, and Zed on this machine. Prompts for your API key, merges the gnosem entry into each config alongside your existing MCP servers, backs up every file it touches. Then restart the affected clients.

Prefer to configure manually? Add this to your client's MCP config:

{ "mcpServers": { "gnosem": { "url": "https://gnosem.dev/mcp", "headers": { "Authorization": "Bearer gn_your_key_here" } } } }

After restart, your assistant sees 5 tools: memory_write, memory_search, memory_list, memory_forget, memory_supersede.

3. Try a write + search from any client

curl -sX POST https://gnosem.dev/mcp
-H "Authorization: Bearer gn_your_key"
-H "Content-Type: application/json"
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"memory_write","arguments":{"content":"I prefer Postgres over MongoDB for greenfield work.","written_by":"manual","tags":["preference","stack"]}}}'

Prototype caveats: single-user memory (no team sharing yet). No client-side encryption yet — memories are stored server-side in plaintext, encrypted at rest by Cloudflare. Semantic search uses Workers AI (BGE-base-en-v1.5, 768-dim). Rate limits: Cloudflare's default DDoS controls only.

Built on Cloudflare Workers + D1 + Vectorize + Workers AI.