Vercel

Your production traffic's traces, pushed into the graph.

One command provisions a Vercel Log Drain that forwards your traffic's traces to Neat's OTLP receiver — no polling, no restart. Neat sees the routes and edge functions production actually hits.

Paste it into your coding agent — it runs the full setup with your credentials.

Set up NEAT's Vercel connector so the graph sees the routes and functions my production traffic actually hits. Vercel is a PUSH connector: `connector add` provisions a Vercel Log Drain that forwards traces to my NEAT daemon — there is no polling. Steps:
1. Make sure my daemon's OTLP receiver (/v1/traces) is reachable FROM Vercel — i.e. the daemon is on a public URL or a tunnel (cloudflared/ngrok). Note that URL.
2. Gather a Vercel API TOKEN (vercel.com/account/tokens); export as VERCEL_TOKEN, don't paste inline.
3. Run: npx neat.is connector add vercel --token $VERCEL_TOKEN --endpoint https://<my-daemon>/v1/traces
   (run `npx neat.is connector add vercel --help` to confirm exact flags for team/project scoping.)
4. The drain is live immediately — no restart. Trigger traffic (deploy, or hit a few routes/functions).
5. Verify within a minute: check the graph for OBSERVED route/function activity from Vercel. Report which routes and edge functions show real traffic, and any that diverge from what the repo declares.

how it works

Catch a route that works in one region and 404s in another — before a user files the bug.

Make your daemon's OTLP receiver reachable from Vercel — a public URL or a cloudflared/ngrok tunnel.
Run the connector with your Vercel token and that endpoint. It provisions the Log Drain — live immediately, no restart.
Trigger traffic. Within a minute the routes and edge functions light up as OBSERVED, and any that diverge from the repo stand out.

Other connectors

Supabase The tables and RPCs your app actually queries in production. MongoDB Atlas Every collection your services actually read and write. Railway Services and deploys, no app code required. Cloudflare Which Workers actually run in production, mapped back to your code.