← SkillSafe / Sentry Init

The Sentry init you already ship, read line by line

Paste your Sentry.init — one file or several, one platform or four. Before anything is sent anywhere, this page parses it: every init block, which SDK it belongs to, every option it sets, the shape of the DSN, and about thirty things that are commonly wrong. Then pick a lane.

Each example ships with a saved model run, so you can see a whole lane — audit, privacy, budget or signals — without signing in and without spending a credit.

nothing pasted yet

Paste whole files if you like — imports help identify the SDK. Separate files with a // file: path line.

Drop files in, or Everything is read in your browser. Nothing is uploaded until you run a lane.
Your traffic and your plan — optional, and the only way the budget lane can do arithmetic

These numbers never leave your browser except as part of a run you start. Leave them blank and the budget lane will say the arithmetic could not be done rather than inventing an average.

Paste an init call to price the run.

Runs use the gpt-terra model and cost credits. The prescan above never does.

How this works

  1. One work object. Every lane reads the same box: the init call your project ships. Nothing else is pasted, and the traffic numbers are yours to leave blank.
  2. The prescan is free and it is real. It finds every init block across eight SDK syntaxes, normalises option names so traces_sample_rate and TracesSampleRate are one row, parses the DSN, and applies about thirty lint rules. Nothing leaves your browser to do it.
  3. Credentials are masked here, not there. DSN keys, legacy DSN secrets, auth tokens and email addresses are replaced with placeholders before the text is sent — and the scan itself runs on the masked text, so there is no second path a secret could travel down.
  4. The arithmetic is the browser’s, not the model’s. The monthly event figures come from your rates and your traffic, computed here. The model is told they are authoritative and is asked for the plan, not the numbers.
  5. Every flag gets an answer. The prompt requires the model to confirm, clear or set aside every prescan flag by id, and the reconciliation table shows any it missed.

Derived from five open-source Agent Skills in getsentry/sentry-for-ai: @getsentry/sentry-sdk-setup and @getsentry/sentry-node-sdk (audit), @getsentry/sentry-browser-sdk (privacy), @getsentry/sentry-nextjs-sdk (budget), @getsentry/sentry-feature-setup and @getsentry/sentry-otel-exporter-setup (signals). This app is a derived work and is not affiliated with those authors or with Sentry.