The AI-native knowledge base for Zed

Zed is fast inside the file. Hjarni is where the project context lives. The assistant panel reads both.

Free to start. No credit card required.

What this unlocks

Workflows that actually use your context

Give the assistant context the repo cannot answer

"Why did we move off gRPC for billing?" Zed's assistant searches Hjarni for the ADR, reads it, and answers from your actual decision instead of inventing one.

Edit threads that start with your conventions loaded

Open the assistant panel against a file. It pulls the Hjarni notes for that project before suggesting changes. The diff matches your error handling and naming on the first try.

Capture the lesson without leaving the editor

After a tricky refactor in Zed, tell the assistant to save a Hjarni note describing the new pattern, tag it for the team, and link it from the related ADR. Next week's session reads it back.

One knowledge base across every editor you use

Zed today, Cursor for a side project tomorrow. The notes do not move with the editor: they live in Hjarni and every MCP-aware editor reads the same set.

Pair-program without re-pasting the architecture

Stop dropping the same 'how the services fit together' paragraph into every prompt. Write it once as a Hjarni note. The assistant reads it on demand instead of guessing.

Setup

Connect Zed in about two minutes

1
Sign up for Hjarni and write a few notes describing how the codebase is meant to fit together.
2
In Zed, open the command palette and run 'zed: open settings' to edit settings.json.
3
Add the context_servers block above. The mcp-remote proxy bridges Hjarni's HTTP endpoint to stdio, which is the shape Zed's custom server entry expects across versions.
4
Save the file. Zed loads the new server and opens a browser tab for the Hjarni OAuth flow on first use.
5
Open the assistant panel and ask it to list your Hjarni notes. The Hjarni tools should appear in the panel's tool list.

~/.config/zed/settings.json (context_servers)

{
  "context_servers": {
    "hjarni": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://hjarni.com/mcp"]
    }
  }
}

The Hjarni MCP endpoint is https://hjarni.com/mcp. The connection uses OAuth, so you sign in with your Hjarni account when prompted. No API key files to manage.

Why a Zed MCP knowledge base sits outside the editor

Zed is excellent at reading the open file and threading edits across it. It is not designed to read the design doc in someone's Notion, the runbook a teammate keeps in Apple Notes, or the decision buried in a Linear comment. Hjarni is one place all of that can land, with MCP search built in.

The assistant's planning step is the highest-leverage moment to inject context. If it starts with the wrong assumptions, the diff is wrong. Folder-level AI instructions in Hjarni let you steer the assistant per project: 'in the Payments folder, never propose schema changes without an ADR'.

Zed is one editor. Engineers use more than one. Hjarni stays put while editors change underneath it, so the notes you write today are still useful when you switch editors next year or when a new teammate joins.

Common questions

Questions before you connect Zed

Where does the Zed MCP config live?

In settings.json. Open the command palette and run "zed: open settings", then add a context_servers block. The same file holds your editor settings, so you can version it alongside your dotfiles.

Does Zed support remote (HTTP) MCP servers directly?

Zed's MCP support has evolved across versions. The most reliable cross-version setup is to launch the mcp-remote npm package as a stdio bridge to Hjarni's HTTP endpoint, which is the snippet shown above. If your Zed build supports direct HTTP MCP, you can swap in a url-only entry instead.

What does Hjarni give Zed that the assistant cannot read from the workspace?

Zed reads the open file and the workspace. It does not read decisions stored in Linear, runbooks in Notion, or the customer interview you wrote in Apple Notes. Hjarni gives the assistant one MCP-backed source for that material.

Can I share a Hjarni knowledge base with my team on Zed?

Yes. Hjarni's team plan gives each engineer their own Zed connection to the same shared folders. Assistant sessions on each machine then read from the same set of notes.

Does Zed need an API token to use Hjarni?

No. mcp-remote opens a browser tab for OAuth on first run and stores the token locally. You sign in once with your Hjarni account, no key files to manage.

Give Zed a memory

The session starts where the last one ended. Write notes once. Zed reads them across every conversation.

Get started free See pricing

More integrations

Hjarni for Claude Hjarni for ChatGPT Hjarni for Cursor Hjarni for Claude Code Hjarni for Windsurf Hjarni for Cline Hjarni for GitHub Copilot Hjarni for Continue.dev Hjarni for Aider Hjarni for Open WebUI Hjarni for Raycast Hjarni for Alfred Hjarni for Obsidian

Write once. You both remember.

Free to start. No credit card required.

Give your AI a memory

Works with Claude and ChatGPT today. Gemini coming soon.