RerunLabChoose an endpoint.
Add a URL and, if needed, a scoped API credential.
GET /jobs/export_123
Polling, handled.
Turn API responses
into events.
Watch a JSON endpoint for changes or a condition. RerunLab handles the repeated requests and delivers the result to your webhook or event inbox.
For developers and agents. No polling worker to maintain.
EXPORT JOB WATCH01 / condition
GETdemo source / jobs / exportJSON
{
"id": "export_123",
"status": "queued"
}/statusequals"completed"
- ✓Baseline capturedRemember the first successful value.queued
- ✓Unchanged checkSame value. No event.queued
- ○Waiting for the conditionComplete the job to see the event.
✓ Event stored · Simulated webhook received
{"type":"watch.matched",
"data":{"before":"queued",
"after":"completed"}}Simulated example — no external request.
HTTPS + JSONDeterministic rulesSigned webhooksDurable event inbox
A SMALL PRIMITIVE. LESS PLUMBING.
An endpoint. A rule.
Something you can act on.
Tell us what matters.
Watch a field for changes or wait for a condition.
/status == "completed"
Get an event.
Receive a signed webhook or retrieve it from your inbox.
watch.matched → your workflow
WHEN “CHECK AGAIN LATER” BECOMES A FEATURE
Keep the work moving.
Wait for a job to finish.
Monitor an export or background task, notify once, then stop.
Add events to a polling-only API.
Integrate a system that has the data you need but no suitable webhook.
Let an agent check back later.
Register a watch, end the session, and consume the event when your workflow resumes.
Watch the field,
skip the noise.
Select a JSON value so unrelated response fields do not trigger events.
Delivery that survives
a failed request.
Signed webhooks, bounded retries, delivery history, and replay for retained events.
A record you
can inspect.
See successful observations, source errors, detected changes, and delivery attempts in one place.
Stop when
the work is done.
Expiring watches and one-time conditions keep temporary monitoring temporary.
BUILT TO FIT YOUR WORKFLOW
One watch.
Dashboard, API,
CLI, or agent.
Create and manage the same durable watch through whichever interface fits your work.
Read the agent quickstart →A human-provided API key is required. Automatic agent wake-up depends on your receiving workflow.
Create a demo job in your workspace, then set its URL and project ID.
curl "$RERUNLAB_BASE_URL/v1/watches" \
-H "Authorization: Bearer $RERUNLAB_API_KEY" \
-H 'Content-Type: application/json' \
-H 'Idempotency-Key: demo-export-watch' \
--data @watch.jsonFrom the local source checkout; no npm package publication is implied.
bin/rerunlab auth login
bin/rerunlab watches create --file watch.json
bin/rerunlab events follow \
--cursor-file .rerunlab-cursorCreate a RerunLab watch for this job's status. Check every five minutes for up to six hours. Store the result in the inbox and stop after it completes.
Five-minute checks require a paid plan. Free watches have a ten-minute minimum. The agent must respect the workspace’s actual allowance.
Events from polling, not magic.
RerunLab still makes requests to the source API. Detection depends on your interval, and changes between checks can be missed. Upstream limits and charges still apply.
PREDICTABLE BY DESIGN
Start free. Grow deliberately.
Free
$0/ month
10,000 included checks · 5 enabled watches
10-minute minimum interval · 3-day event history
Start free →Builder
$19/ month
100,000 included checks · 50 enabled watches
1-minute minimum interval · 7-day event history
View Builder →No automatic overages. At your limit, polling pauses. Existing events and deliveries remain available.
ONE LESS WORKER TO MAINTAIN