The words
The jobs
Every piece of work runs as a background job. It shows up on Agent as a conversation, and on Notifications once it settles.How the work chains
You rarely start a job by hand. Each one starts the next:1
Connecting a repository starts Flow setup
Flow setup reads the code, not the documents, so it runs even before any documentation is linked.
2
A source sync that changed something starts a Document scan
Sources sync when they are added, when you press Sync now, and on their own schedule. See Sources.
3
Flow setup starts Flow generation
It does so when the recipe held and the repository reads documents.
4
Flow generation starts a Flow run
It does so once it has stored tests. This first run is the repository’s baseline.
5
A scan that changed the corpus starts work for every repository it affects
When no conflict is open, each repository whose slice moved gets Flow setup if it was never set up, or Flow generation if it was.
Drivers
A test drives the program through one of its surfaces:
A flow that needs a TUI, library, desktop or mobile driver waits for one; those drivers do not exist yet. Flows can be filtered by driver.
Flow statuses
Every flow wears one of five statuses:Where to read results
- Flows lists every flow of every repository. Filter by Status, Driver and Repository. Open a flow to see its milestones, their cases and its tests.
- A repository’s Runs tab lists every Flow run, and each run opens with its evidence.
- Home shows flows by status over time, what needs attention, and coverage by area.
- Agent shows each job as a conversation, including why a job stopped.
- Notifications keeps one row per job, from start to how it settled.
Next steps
Document scan
How the workspace’s documents become one curated corpus.
Flow setup
How a repository is prepared to be tested.