Unl — Claude Code MCP server: what it does, how to install
Stores engineering decisions with their underlying rationale and feeds relevant context to coding agents before each step. Keeps your tools aligned on project intent.
claude mcp add --transport http unl https://api.unlimitless.ai/mcpWhat Unl does
Unl tracks technical choices and the context behind them across your coding workflow. When your agent prepares to write code, the server injects the specific guidelines that affect the immediate task. It avoids repeating project history in every prompt. You confirm every new rule.
Client libraries, the CLI, and the decision format are open source under the MIT license. The retrieval engine itself runs remotely at api.unlimitless.ai, meaning your project decisions pass through hosted infrastructure. Integration relies on OAuth verification instead of pasting raw API secrets into client configuration files.
What it helps with
- persist architectural decisions along with their rationale
- share decision context across Claude Code, Cursor, and Codex
- feed relevant project constraints to models before they modify files
- review agent-proposed rules prior to saving them
Example prompts for the agent
- Check Unl for the decisions we recorded about background job processing.
- Show me why we chose synchronous exports in this service.
- Draft a new decision entry for our auth flow and send it to Unl.
How to install Unl
Platform CLI: remote server
claude mcp add --transport http unl https://api.unlimitless.ai/mcpPlatform CLI: remote server
codex mcp add unl --url https://api.unlimitless.ai/mcpConfig file
{
"mcpServers": {
"unl": {
"url": "https://api.unlimitless.ai/mcp"
}
}
}Config file
{
"servers": {
"unl": {
"type": "http",
"url": "https://api.unlimitless.ai/mcp"
}
}
}Platform CLI: remote server
gemini mcp add -t http unl https://api.unlimitless.ai/mcpStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Unl
How to install Unl in Claude Code?
Run claude mcp add --transport http unl https://api.unlimitless.ai/mcp in your terminal. After adding the server, open /mcp, highlight unl, and complete the OAuth login.
What does Unl do?
It keeps a record of engineering choices alongside the reasoning behind them. Whenever an agent works on a task, Unl surfaces the decisions that directly apply to that code.
Where is decision data processed?
The CLI tools and schemas are MIT-licensed, but the core ranking engine runs on api.unlimitless.ai. Connecting a client uses OAuth without pasting personal tokens into local configs.
