Example processes
Six of the example processes from minctrl.com/flows, imported into your company and run end to end — import, start a run, it parks at a human gate, sign, outcome.
Minctrl does not ship processes into your company. The flows on
minctrl.com/flows are examples: each has a
minctrl.process/v1 file at https://minctrl.com/examples/<slug>.process.json that the
import endpoint accepts unchanged. Once imported, an example is an ordinary process in your
company — versioned, editable on the canvas, and run like one you built.
Each recipe below is import → start a run → it parks at a human gate → sign → outcome.
Every step is a curl you can paste.
Setup for every recipe
Register once, then import the example the recipe uses (replace <slug>):
TOKEN=$(curl -s -X POST "$API/auth/register" \
-H "Content-Type: application/json" \
-d '{"email":"[email protected]","password":"<your-password>","company_name":"Acme Ops"}' \
| jq -r .token)
SLUG=<slug>
curl -s -o "$SLUG.process.json" "https://minctrl.com/examples/$SLUG.process.json"
jq -n --slurpfile doc "$SLUG.process.json" '{document: $doc[0]}' \
| curl -s -X POST "$API/process-templates/import" \
-H "Authorization: Bearer $TOKEN" \
-H "Content-Type: application/json" \
--data-binary @- | jq '{slug, version, executable_violations}'The import is a draft version. Every run in the recipes names the process by its slug
with "draft": true, which starts the published version: submit the import for review
and have an owner or admin publish it — with separation of duties on, one who did not import
it; working alone, turn it off and publish it yourself (see
Publishing processes). Until then you, the importer,
can run a recipe as a shadow test run by adding "draft_version": 1, "shadow": true.
Conventions shared by every recipe:
gateon resume is the step-id the run parked on (trace.parked_step, e.g.bsa-gate,signoff) or itsparked_atlabel (e.g.gate:bsa-officer-signoff). A value that names neither leaves the run parked where it was.- Signing needs the manager role or above. With separation of duties on (the default), one person cannot sign two different gates of the same run.
- Irreversible tools stay shadowed (no real call) unless a connector opted in with
reversible: true— see Connect a step.
How the examples route without your systems
With no connectors bound, steps run on a built-in heuristic executor. At a branch it takes
the edge whose label contains your risk input, otherwise the first edge. After a gate, the
verdict picks the branch: the branch label it names, otherwise by meaning (approved takes
the approval branch, denied the denial or rework branch); a verdict that resolves to no
branch leaves the run parked. The inputs in each recipe are chosen to reach the gate
described, and the outcomes are what this simulated run does. With your connectors bound,
the branches follow your systems' answers.
AML alert triage
Example aml — screens and scores; a high-risk party parks at
bsa-gate for the BSA/AML Officer.
Prior authorization
Example prior-auth — parks at signoff for a medical director.
Debt collection
Example collections — parks at signoff for a manager.
Chargeback representment
Example chargeback-representment — parks at a fight-vs-refund gate.
SIM-swap fraud defense
Example sim-swap-fraud-defense — a flagged change parks at
signoff for a fraud analyst, then completes.
Provider credentialing
Example credentialing — parks at committee-signoff, then again at
enrollment-approval before payer enrollment.
Self-hosting — Kubernetes, Hetzner and the data region
Run Minctrl on your own infrastructure with the Helm chart or the Hetzner Terraform, keep its data in one region, and know exactly what is tested and what stays yours to do.
AML alert triage
Onboard and screen a customer with the aml example, park at the BSA officer gate, sign it, and keep the run monitoring — every SAR and OFAC action stays shadowed until a human signs.