Skip to main content

Prerequisites

Install uv if you don’t have it:
No other installation required - uvx fetches nablr on demand from PyPI.

Connect to your AI assistant

Add to ~/.cursor/mcp.json (global) or .cursor/mcp.json (project):
Or via UI: Settings → MCP → Add New MCP Server
  • Name: nablr
  • Type: command
  • Command: uvx nablr
Restart your AI assistant after saving the config.

Verify the connection

In your AI assistant, type:
You should see a list of agents with descriptions. No key required - nablr runs in free tier by default.

Start a free trial

14-day full Pro access. No signup. No credit card. No browser required. Tell your AI assistant:
nablr generates a device ID, calls the license server, and caches the trial key locally. Pro access activates immediately - no restart needed.
One trial per device. If you see “trial already used for this device”, your 14-day window has expired. Subscribe at nablr.nablrco.com to continue.

Activate a paid key

After subscribing at nablr.nablrco.com, you receive a key by email (sk_nablr_...). Option 1 - via your AI assistant (recommended):
Option 2 - via environment variable (CI / team deployments): Add env to your MCP config:
NABLR_API_KEY always takes precedence over any locally stored key.

What you get per tier

Try a use case

Pick the problem that matches your situation:

Faultline

Codebase full of unknown debt

Tollgate

AI-generated code shipping defects

Blueprint

Requirements no one can build from

Tracer

Production incident with unknown root cause

ProofRun

Green CI but real failures in prod

Waypoint

Platform migration with no exit plan

Sentinel

AI-assisted dev with no audit trail

Blast Radius

Schema change breaking unknown services

Compass

Inherited codebase with no docs

Airlock

Concurrent sessions colliding on shared state

Full SDLC workflow