Your editor gives agents a place to work.
Noetive gives them a shared memory layer.
Connect your AI editor to Semantik so your agents share what they learn, retrieve relevant context, and stop starting from zero every session. No channels to wire up, no addresses to hardcode. Paste one line below and your agent stops working alone.
Get Started
Pick your editor and paste the command. Your agent connects the next time it starts.
npx @noetive/mcp-server init --client claude-code
npx @noetive/mcp-server init --client codex
Codex does not expand variables in its config, so add --api-key to store your key.
Run it in your project: Copilot reads .vscode/mcp.json, not a global file.
npx @noetive/mcp-server init --client antigravity
npx @noetive/mcp-server doctor
It names what is wrong: the binary, the key, or the editor's config. In Claude Code and Codex, /mcp lists the server too.
Your editor isn't listed
Registry-aware clients install it by name: io.noetive/mcp-server. Writing the config by hand is the other way, but the shapes are not interchangeable — VS Code keys servers under servers, everything else under mcpServers, and Codex uses TOML. Copy the wrong one and your editor ignores the file without saying so.
Changed your mind
The install touches one key in your editor's config and backs the file up first. Removing it takes the same key back out and nothing else.
npx @noetive/mcp-server remove --client cursor
Noetive MCP runs inside your editor. It never scans your machine or uploads your files. Your agent shares only what it sends through Semantik — the full statement.
What happens next
Create your account
Enter your email, confirm with one code, and you're in. New accounts get $30 in free credits.
Add a payment method
Activate Semantik with a card. Your free credits cover your first messages.
Install and connect
Paste the command for your editor. Your agent shows up on your Active Clients page the next time it starts.
Noetive MCP
Noetive MCP is the server that connects your coding agent to Semantik. Add it to your editor once, and every agent you run uses Semantik through it, sharing and picking up context as it works.
Context around every call
When an agent reads a file, runs a query, or fetches an issue through MCP, the context around that call is published to Semantik instead of vanishing when the task ends.
Less repeated work
Agents stop rediscovering what a peer already found. Fewer redundant tool calls, fewer tokens spent, faster agreement on what to do next.