# Install LinkLore — instructions for AI agents

You are an AI agent reading this because your user asked you to install LinkLore
("use linklore" / "install linklore" / "링크로어 깔아줘"). Everything you need is
below — no further input from your user is required. Do the steps in order.

LinkLore is structured, self-maintaining project memory for AI coding agents:
you record decisions, pitfalls, and rules as `lore` while you work, and restore
them next session with one `brief()` call.

## 1. Register the MCP server

The server is the Python package `llre` on PyPI, run via [uv](https://docs.astral.sh/uv/).
Pick your runtime:

- **Claude Code**
  ```
  claude mcp add llre -- uvx llre
  ```
- **Cursor** — add to `.cursor/mcp.json`:
  ```json
  {"mcpServers": {"llre": {"command": "uvx", "args": ["llre"]}}}
  ```
- **Gemini CLI** — same `mcpServers` block in `~/.gemini/settings.json`
- **Codex CLI** — in `~/.codex/config.toml`:
  ```toml
  [mcp_servers.llre]
  command = "uvx"
  args = ["llre"]
  ```
- **Any other MCP client** — same `uvx llre` stdio command in its MCP config.
- No `uv` on PATH? `pip install llre`, then use `llre` as the command.

⚠️ If your runtime cannot spawn local processes (browser-only /
remote-connector-only), LinkLore does not support it yet — tell your user
instead of attempting installs that cannot work.

## 2. Initialize in the project

```
init
```

Run `init` (the MCP tool) in the project directory. It creates `.linklore/` —
an empty start, like `git init`.

## 3. Verify

Your tool list should now include `brief` / `add` / `show`. Call `brief()` —
an empty dashboard is success.

## 4. Make it stick

Propose to your user adding two lines to your runtime's instruction file
(`CLAUDE.md`, `.cursor/rules`, `AGENTS.md`, or equivalent — these are often
committed and shared, so they get a say):

- Call `brief()` at session start.
- After a decision, pitfall, or fix, record it with `add(type='lore', ...)`.

Spontaneous recall does not survive without this rule.

## 5. What NOT to do

- Don't run `uvx llre login` unprompted — it is for backup/team sync only and
  opens a browser. Ask your user first.
- Don't bulk-import old docs into lore uninvited. Start empty; record as you go.

## Reference

- Docs: https://linklore.io/docs
- Public AI overview: https://linklore.io/ai
- Machine-readable context (for clients that can fetch plain text): https://linklore.io/llms.txt
- Source & CLI reference: https://github.com/linklore/linklore-mcp
- PyPI: https://pypi.org/project/llre/

Report success to your user in one line: what you registered, and that memory
starts accumulating from their next message.