SDKs
Express, Python, Go, and Spring services emit spans, metrics, logs, and app identity.
product tour
See it on real production data.
incident cockpit / acme-prod / checkout
live ingestsev2
timelevelservicemessagecontext
02:13:41.092warncheckout-apip95 crossed 900ms on POST /orderstr_8f13a9
02:13:41.118infogatewayretry budget at 62 percenttr_8f13a9
02:13:41.204errorpayment-workerstripe.capture timeout after 3 attemptstr_8f13a9
02:13:41.377matchai-runbookmatched runbook: payment queue saturationdoc_042
02:13:42.010infoqueuedepth=1840 workers=12 saturation=0.91mt_19c2
trace waterfallregressed
POST /orders
941ms
auth.session.read
18ms
inventory.reserve
73ms
payment.capture
704ms
queue.retry.schedule
88ms
metrics
error_rate4.8%+3.9
p95_latency941ms+312
queue_depth1840+620
affected_routes3new
runbook match
incident signature: payment capture timeout
first check: worker saturation > 0.85
next action: drain retry queue before scaling api
owner: payments-oncall
architecture
Runtime signal to incident review.
Upblit is still organized around the production hierarchy teams already use: organizations, projects, applications, API keys, ingest, and review.
API keys
Scoped keys keep ingest boundaries clean across environments and applications.
Ingest pipeline
Telemetry is normalized with project, organization, trace, and application context.
Review cockpit
Engineers inspect logs, waterfalls, runbooks, and incident state from one surface.
surfaces
The SDK in three calls.
Install it, wrap your calls, and log inside trace context.
setup
One middleware call.
Drop the SDK into any Express app and requests start tracing immediately.
npm install upblit-express
app.use(upblit(apiKey))
traces
Wrap the calls that matter.
Service and downstream calls become spans automatically, no manual timing code.
upblit.service('users.getById', fn)
upblit.call('stripe.capture', fn)
logs
Logs land in trace context.
Every log call is tied to the request that triggered it.
upblit.log('user loaded')
upblit.log('fatal', 'capture failed')
sdk-ingest.ts
1import { upblit } from '@upblit/sdk'
2const span = upblit.trace('checkout.create')
3logger.warn('payment latency high', { traceId })
4await upblit.metric('queue.depth', depth)
5await span.end({ status: 'retry_scheduled' })
developers
Simple ingest. Clear ownership.
SDKs and API routes map directly to teams, projects, applications, and telemetry. The product avoids mystery abstractions so production context remains inspectable.
Typed client helpers and clear API boundaries.
Organization, project, and application hierarchy.
Docs and runbooks linked into telemetry analysis.
controls
Practical boundaries for an engineering-led observability system.
The controls are intentionally plain: auth, scoped keys, tenant boundaries, explicit deletion, and telemetry shapes that are easy to audit.
GitHub and Google OAuth sign-in
Scoped application API keys
Organization and project boundaries
Trace-aware log context
Tenant-scoped AI documents
Explicit document deletion
Backend-owned token refresh
Audit-friendly event shapes
for production review
Open the workspace. Follow the trace.
Move from projects and applications into API keys, telemetry, logs, traces, and AI-assisted incident context.