Documentation
Everything you need to wire GlitchReplay into a project, configure it, and pull data back out.
GlitchReplay is Sentry-SDK compatible. If you can configure a Sentry DSN, you can configure GlitchReplay — point any @sentry/* SDK at your project DSN and events flow in. The pages below cover everything else: replay, source maps, alerts, the REST API, and the security signal we extract from the same event stream.
Sign up, create a project, get a DSN, and verify your first event in five minutes.
Drop-in config for @sentry/nextjs, @sentry/react, @sentry/node, and the browser bundle.
Upload .js.map files at deploy so stack frames render with original file paths.
Capture the user's last 30 seconds of DOM activity before each error.
CSP violations, scanner probes, auth spikes — categorized inline with your errors.
Detect ad-blockers and verify failing image / script / link loads before they hit your dashboard.
Email, Slack, and webhook routing for new issues, spikes, and security events.
Server-side rules that strip personal data before it ever lands in storage.
Per-project toggles for which event-context fields are recorded — URL, referrer, release, user identifiers, query-string scrubbing.
Read-only token-authenticated access to projects, issues, and events.
How events are grouped into issues, and how to override grouping when the default isn't right.
Troubleshoot
Diagnostic walk-throughs for the things that go wrong most often. Skim these first when something isn't working — most reports we see come down to one of these checks.
You installed the SDK and triggered an error, but the dashboard is empty. Six causes, in priority order.
Maps uploaded, release tagged, and the dashboard still shows minified frames. The matching contract and how to debug it.
Architecture at a glance
Every GlitchReplay project has a DSN — a Sentry-format URL with a public key and project ID. Your application's SDK posts events to POST /api/{project_id}/envelope/, the ingest worker validates the DSN and rate-limits, then enqueues the raw envelope. A consumer worker parses each event, fingerprints it, scrubs PII, runs the security classifier, and writes to D1 + R2. The dashboard reads back from D1.
The whole pipeline runs on Cloudflare — Workers, Queues, D1, R2, KV. There is no cross-region egress, and there is no per-event price ramp.
Conventions in these docs
- DSN: replace
<DSN>in any snippet with the value from Project → Setup. - Project ID: the
prj_…string at the end of a DSN. Used in API URLs. - API token: a separate
grt_…token from API Tokens; never the DSN. - Cloudflare runtime quirks (Workers, OpenNext) are called out where they matter.