Small UiPath versions of pieces of Evelyn, WorkFusion's Names Sanctions Screening digital worker. Each simulation copies one Evelyn behaviour at toy size, with mock data, so the Evelyn team can try the UiPath building blocks (Maestro Flows, Python coded functions, storage buckets) before the real port.
Why this exists: Evelyn is moving from WorkFusion to UiPath. A simulation shows the UiPath shape of one Evelyn piece without touching Evelyn's code, its data or a shared tenant. When the port starts, each simulation is a worked example of one piece.
| Folder | Evelyn piece it simulates | UiPath pieces it shows | Status |
|---|---|---|---|
| CsvProcessor Simulation (run it) | CSV ingestion: CsvTaskProcessor and CsvSchemaProducerProcessor, and the CSV BP calling the main BP |
A Flow calling a Flow, a contract shared by two function projects, a storage bucket | Verified locally; not run on a tenant yet |
| Configuration Simulation (run it) | The configuration: configuration.json (the form), the saved values, INPUT_TYPE, and PROVIDER_ID picking a provider |
Settings loaded once per run in three layers, a Switch with one node per provider, an inout global, a Credential asset for the secret |
Verified locally; not run on a tenant yet |
| Dependency Simulation (run it) | Shipping a library: the fircosoft-trust and address-lookup-jnw jars, declared in a pom.xml and bundled into Evelyn's package |
A library in its own project, built into a wheel, vendored inside a coded function, pinned in pyproject.toml and uv.lock |
Verified locally; not run on a tenant yet |
| Datastore Simulation (run it) | Two related datastore tables: alert_analytics and hit_analytics, one alert with zero or many hits, linked by a text column with no foreign key |
Two Data Fabric entities, customers and orders: a primary key, a unique business key, a RELATIONSHIP foreign key, a lookup before insert, a LEFT JOIN |
Verified locally; not run on a tenant yet |
Each folder stands on its own, with two guides:
- HOW_TO_RUN.md runs what's there: the tests, a local run, the tenant run, then a deployed run started with curl or Postman. It also says where to find the run, the bucket and the asset in the browser. Start here to see it work.
- README.md explains what it simulates, and builds it from an empty folder by commands. It also has the expected result and the troubleshooting.
| Tool | For |
|---|---|
uv |
Python 3.11 projects and their tests |
The uip CLI, logged in (uip login status) |
Solution and Flow commands |
jq |
The JSON edits in the build steps |
unzip |
Looking inside a package (Dependency Simulation) |
| A UiPath tenant | Only levels 3 and 4 of each HOW_TO_RUN.md |
curl, or Postman |
Level 4: starting a deployed Flow over HTTP |
- Mock data only. No real people, no customer data, no secrets. Names are placeholders such as Alice and Bob, John Doe and Foo Bar, or tech names with their first letters swapped (Gill Bates).
- No tokens in git.
.gitignorekeeps.env,.venv/and caches out. Never runuipath authinside a project folder: it writes a token into.env, anduip solution packships.env. Useuip login. - Rebuildable. Each README rebuilds its simulation from an empty folder by commands and says what has and hasn't been verified.
- Unique names. UiPath names are tenant-wide, so each simulation uses its own function, Flow and bucket names.
AGENTS.md (and CLAUDE.md, which loads it) tells an AI assistant how this repo is laid out and how to add the next simulation. Before it designs anything, it asks for your local checkouts of Evelyn's code and its names classification model, and warns you if it can't find them.