Skip to main content

Keep a pulseon what you ship.

With AI writing so much of your code, you can't test every path, and some bugs only show up in production. What matters is fixing them fast. Patchly catches the crash, explains the root cause in plain English, and opens a draft pull request with the fix.

Your app, drawn as a small ship, sails from localhost into production, hits a TypeError reef and starts leaking. Patchly, a small lifeboat, sails up, finds the cause, patches the hull and sails off while your app carries on.4971812localhostopen water: productionTypeError: cart.items[0] is undefinedYour appRoot cause: the cart can be emptyDraft PR opened · CI passingMerged · back on coursePatchly

1 · Set up

Set up with one prompt.

Your coding agent built the app, so it can add monitoring too. Create a project, copy the prompt and paste it into Claude Code, Cursor, Codex, Lovable, Bolt or any other coding agent.

  • Browser and server capture, with source maps so stacks point at your code, not the bundle.
  • Release tagging from the Vercel commit SHA, so you know which deploy broke.
  • Automatic environments: local runs report as development, deployed runs as production.

The real prompt, placeholders in brackets

Install Patchly for this app.

Project ID: <your project id>
PATCHLY_DSN=<your DSN>
PATCHLY_AGENT_TOKEN=<your agent token>

Please:
1. Install patchly@latest
2. Add browser and server init
3. Use the Vercel commit SHA as the release when available
4. Upload source maps if this app builds them
5. Leave environment auto-detected so local runs report as development and deployed runs as production
6. Keep the integration minimal and production-safe

Full setup guide: https://patchly.cc/SKILLS.md

2 · Diagnose

Know when a bug only happens in production.

Patchly keeps local and production events apart and lines them up for the same bug. If it only happens in prod, it tells you so, and shows exactly which release, browser, runtime or URL differs.

The cause is usually environment variables, real user data or the production build. Seeing what differs tells you where to start.

Local vs production

example

Production

212

events

Local

0

events

Release
Only in production: a1b2c3d
Only locally: working tree
Runtime
Only in production: edge
Only locally: node 22
URL
Only in production: /checkout?new=1
Only locally: /checkout
Only seen in production (212 events). It hasn't happened locally, so look at what differs in prod.

3 · Fix

Fixes arrive as drafts. You stay in control.

Patchly writes the fix against your real source and opens it as a draft pull request labelled untested. Your CI runs on it, you review the diff, and you decide whether it ships.

Never merged for you

Patchly opens the draft and stops there. Marking it ready and merging always takes a person.

Never touches your production database

Patchly reads error events and, through the GitHub App, your source. It has no connection to your database, so it cannot change your data.

Honest when unsure

If there isn't enough context to be confident, the repair note says so instead of inventing file paths or diffs.

Also included

Security alerts for code you actually run.

AI coding agents install a lot of dependencies, and many never load in production. Patchly checks your lockfile against public advisories and ranks the ones attackers are exploiting, or your app actually executes, first.

Rolling out now as the Security tab in your project dashboard.

Known exploited
Listed in CISA's KEV catalogue: attackers are using it right now.
Likely to be exploited
A high EPSS score: the probability of exploitation in the next 30 days.
Seen at runtime
The package shows up in your real stack traces, not just a lockfile.

Ship fast. Fix faster.

Free for 10,000 events a month. Paid plans from $10/month, with no per-seat fees.

Report a bug
    Patchly — Keep a pulse on what you ship