# The AI-native knowledge base for Cline

The Cline MCP server for long-term memory. Cline plans, edits, and runs commands. Hjarni is where the project context lives.

## Workflows that actually use your context

1. Give Cline long-term memory before plan mode runs

   Cline reads the relevant Hjarni notes while planning. Instead of inventing conventions, it lays out a plan that respects your tests, your error handling, and your past refactors.

2. Hand off a long task cleanly

   At the end of a Cline run, ask it to write a Hjarni note: what changed, what is left, what to test. The next task picks up from that note without a recap.

3. Skip the architecture preamble

   If you keep pasting the same architecture summary into Cline before each task, save it as a Hjarni note and tag the right folder. Cline reads it on its own from then on.

4. Make the work visible to humans

   Cline's chat is private to your VS Code window. Hjarni notes are searchable in the browser and shared with your team. Pair the two and the work stops disappearing into a session log.

5. Reuse Cline's output in other tools

   Notes Cline writes back to Hjarni are read by Claude, ChatGPT, and Cursor too. The cleanup work pays off in every other interface.

## Connect Cline in about two minutes

1. Install the Cline extension in VS Code and sign up for Hjarni.
2. Open the Cline panel in VS Code. Click the MCP servers icon at the top of the panel.
3. Choose Add server. Pick remote, name it hjarni, and set the URL to https://hjarni.com/mcp.
4. Cline opens a browser tab for the Hjarni OAuth flow. Sign in and approve.
5. Run a Cline task and confirm the Hjarni tools show up in the tool list before plan mode runs.

```
{
  "mcpServers": {
    "hjarni": {
      "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.

## Why a Cline MCP knowledge base outlasts the memory bank

Cline is open source and moves fast. That is a feature for the agent, but it means the right strategy is not to tie your knowledge to Cline-specific files. Keep notes in Hjarni, point Cline at the MCP server, and survive every Cline update without re-organizing anything.

Cline's memory bank lives in your workspace and is great for short-lived task state. Hjarni complements it: long-lived architecture decisions, runbooks, and team conventions live in Hjarni, while the workspace handles the per-task scratch.

Cline runs commands and edits files autonomously. The agent is faster than you can read. A shared Hjarni folder is how you keep humans, Cline, and other assistants on the same page: write decisions once, all three read them.

Common questions

## Questions before you connect Cline

**How do I add Hjarni to Cline?**

Open the Cline panel in VS Code, click the MCP servers icon, and add Hjarni as a remote MCP server. Cline writes the entry to its settings JSON for you.

**Does Cline's plan mode use MCP tools?**

Yes. Cline can search Hjarni during planning, read the relevant notes, and reference them when drafting the plan. The same tools are available in act mode.

**Why pair Cline with a knowledge base at all?**

Cline is excellent at executing tasks once it knows the context. Hjarni is where you write that context down. Together they replace the back-and-forth of pasting architecture notes into every conversation.

**Does Hjarni cost extra on top of Cline?**

No. Cline is free and open source. Hjarni has a free tier with full MCP access. You pay your own model provider for tokens, exactly like Cline already works.

**Can Cline create notes back into Hjarni?**

Yes. Cline can call any of Hjarni's MCP tools: search, read, create, update, link, tag. Useful for saving the outcome of a debugging session as a note.