# The AI-native knowledge base for Continue.dev

Continue.dev is your open-source AI coding assistant. Hjarni is the knowledge base it reads across every project, every model, and every chat.

25 notes free, forever. No credit card required.

## Connect Continue.dev in about two minutes

1. Install Continue.dev in VS Code or JetBrains and sign up for Hjarni.
2. Open ~/.continue/config.yaml. Continue creates the file the first time you open its settings panel.
3. Add the mcpServers entry to the YAML. The streamable-http transport matches Hjarni's remote MCP endpoint.
4. Save the file. Continue reloads the config and opens a browser tab for the Hjarni OAuth flow on first use.
5. In a Continue chat, confirm the Hjarni tools appear in the tool list and ask the assistant to search Hjarni for one of your notes.

```
name: Local Assistant
version: 1.0.0
schema: v1
mcpServers:
  - name: hjarni
    type: streamable-http
    url: 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.

## Workflows that actually use your context

1. Pair-program with your conventions loaded

   Continue reads the relevant Hjarni notes before suggesting code. Edits land closer to your error handling, test naming, and folder layout on the first try.

2. Skip the architecture preamble in every chat

   If you keep pasting the same 'how the services fit together' paragraph, save it once as a Hjarni note. Continue reads it on demand instead of guessing.

3. Capture the lesson while it is fresh

   End a debugging session by asking Continue to write a Hjarni note: what failed, what you tried, the actual fix. The next on-caller's Continue reads it back.

4. Bring your own model, bring your own notes

   Continue lets you pick the model. Hjarni lets you pick the context. Together they replace the bundled-AI-plus-bundled-context model that locks you in.

5. One knowledge base across editors

   Continue runs in VS Code and JetBrains. Hjarni is editor-agnostic, so the notes you write in IntelliJ today are the ones VS Code reads tomorrow.

## Why Continue.dev pairs naturally with an MCP knowledge base

Continue's design choice is to keep the editor open and the model swappable. That is the right choice. The matching choice on the knowledge side is to keep the notes outside the editor and outside any one model, so the same notes serve Continue today and any other assistant tomorrow. Hjarni is that layer.

Continue's context providers handle in-repo files well. They are not designed for the cross-project, cross-editor knowledge: ADRs, runbooks, incident notes, customer feedback. Hjarni gives Continue an MCP-backed source for that material, alongside the context providers you already use.

Open source on both sides matters. Continue is open source. Hjarni's notes are plain Markdown, exportable as a ZIP at any time. If you leave either product, the work goes with you.

Common questions

## Questions before you connect Continue.dev

**Where does Continue's MCP config live?**

In ~/.continue/config.yaml under the mcpServers key. Continue creates the file the first time you open its settings panel, and it watches the file for changes.

**Does Continue work in both VS Code and JetBrains?**

Yes. The same config.yaml is read by both. If you use Continue in IntelliJ at work and VS Code at home, the Hjarni connection moves with you.

**Does Continue store my notes locally?**

No. Continue calls Hjarni over MCP when the assistant needs context. The notes stay in Hjarni and are fetched at the moment the model asks for them.

**Is Hjarni free if I am using Continue?**

Yes. Hjarni's free tier includes full MCP access. You bring your own model through Continue, exactly as Continue already works. There is no per-message fee on the Hjarni side.

**Can Continue write notes back to Hjarni?**

Yes. The Hjarni MCP server exposes tools for searching, reading, creating, and updating notes. Continue can call any of them, so you can ask the assistant to save the outcome of a debugging session as a Hjarni note.