CEDX Scripts · Developer & low-code
Scripts runs the estate's scheduled automation: cron, interval and event triggers, agent-initiated runs that queue for human approval before touching production secrets, and every run priced in minutes and cents.
The run table shows its arithmetic: compact-audit succeeded in 2.4 minutes for $0.10; allocate-stock failed in 20.2 seconds for $0.03. Automation you cannot price is automation you cannot govern.
Runs on demo data — the cost cards are labelled estimates on screen.
What it is
23 runs sit in the awaiting-approval queue — agent-triggered work that touches production secrets. ledger-post, ar-dunning and fx-rate-pull each hold an Approve and a Reject button, with the initiating agent named. Automation proposes; a person disposes.
320 runs with status, duration, cost, trigger and actor. run_0eadcb13 — compact-audit, agent-triggered, approved — succeeded in 2.4 minutes at $0.10. notify-digest failed in 2.8 minutes at $0.20. Success and failure both get a bill.
The Logs view covers 47 distinct scripts across 240 lines, and the overview's secret board names the exposure: sec_billing_settle_db at 312 days, sec_legacy_v1_token at 401, both at risk 95. Thirteen credentials sit past rotation SLA — as an alert, not an audit finding.
Product tour
Not a mockup and not a concept deck. This is what opens at /app/scripts.
01 — Overview
320 runs at 62.8% success against a ≥95% target — the ring shows 63%, not a rounded-up fiction. Alerts arrive root-caused: a retry storm on partner-orbit-sync at 11.4% fail with the retry budget often exhausted, and a ledger-post timeout cascade under event fan-out.
02 — Schedules
36 schedules with kind, expression, next fire and fire counts. on-order-settled has fired 2,100 times on event:order.settled; the edi-webhook listens on webhook:/hooks/edi. Enabled state is a column, and 30 of 36 are firing.
03 — Runs
Runs stream in with the actor column doing the governance: Agent rows carry approved chips, Human rows read Manual, and retry-triggered runs link back to their failure. The sample's bill is $63.87 across 1,011 minutes.
04 — Logs
240 lines across error, warn, info and debug, searchable by message, script, run or step. notify-digest errors on “cleanup done · step”; price-sync warns “timeout approaching” before the run is killed. The log line is where the run table's red row sends you.
Who runs it
Roles, not references. We have no named customers yet, so nobody in these photographs is quoted, credited or claimed as one.
Owns the schedules and the scripts behind them — and reads the retry storm on partner-orbit-sync as a design problem, not bad luck.
36 schedules · 2,100 fires on one
Works the failed and timeout rows: 40 failed, 28 timed out in the sample, each with a log line and a retry button rather than a shrug.
62.8% success · target ≥95%
The human in the loop for the 23 queued agent runs — approve or reject, with the secret scope and the initiating assistant on the card.
23 awaiting approvalThe shape of it
Every figure below is legible in the captures above. Nothing here is a projection of your estate — it is the state of the demo data.
How it runs
Work registers as cron, interval or event — on-order-settled listens for event:order.settled and has fired 2,100 times.
Agent-initiated runs that touch production secrets queue for a human: 23 waiting, each with Approve and Reject on the card.
Execution bills by the minute — the sample's $63.87 across 1,011 minutes — with duration, cost and actor on the row.
Logs stream per script, and secrets past rotation SLA surface on the overview: 13 credentials at risk ≥55, the oldest 401 days.
One record
Scripts is not a task runner pointed at webhooks. Its triggers, secrets and runtimes are the estate's own.
Event-driven schedules subscribe to estate events — order.settled and invoice.paid are the streams Scripts listens to.
CEDX FunctionsScript runs execute on the same metered runtime — duration billed, cold starts costed, failures logged.
CEDX DeployA script ships through the same environments and risk gates as any other deploy before its schedule goes live.
CEDX TablesThe agent runs writing into records land here — fill bars and agent-run states are Tables' side of the same work.
Limits
Finding this out on the third call is worse for you than reading it here, and worse for us.
Start
Open it right now
Your first schedules
Scripts with the rest of it
Questions
Yes. Every screenshot is a capture of the running build and you can open the same build at /app/scripts. It runs on demo data from the Northline Systems demo workspace.
The approval queue. Agent-triggered runs that touch production secrets hold for a human — 23 are queued in the demo workspace — with the initiating assistant and the environment on the card. Nothing auto-approves.
It is priced per run in minutes and cents: compact-audit at $0.10 for 2.4 minutes, email-bounce at $0.40 for 5.1. The month-to-date card reads $3.3K and is labelled an estimate on screen — it is not a price list.
Cron, interval and event. The demo workspace runs 20 cron, 7 interval and 9 event-driven schedules, with expressions like event:order.settled and webhook:/hooks/edi readable on the row.
No certification has been issued. What we can evidence about hosting, encryption, tenant isolation and production access is written up on the security page.
Live build, demo data, no card. Then find the schedule in your estate that nobody has reviewed this year.