These limits are built into Watcher Community Alpha. They are bounds, not performance guarantees.
Runs
| Item | Limit |
|---|
| Active managed runs | 1 at a time |
| Active Playwright Tool sessions | 1 at a time; UI Mode cannot run alongside a managed run |
| Planned test collection | Times out after 15 seconds |
| Workers override | 1–64 |
| Retries override | 0–10 |
| Selected projects | Up to 32 |
Live output
| Item | Limit |
|---|
| Lines retained in the browser view | 1,000 |
| Lines retained on the server for reconnecting browsers | 500 |
| Length of a single displayed line | 64 KiB, then truncated with [Watcher: log line truncated] |
History
| Item | Limit |
|---|
| Runs shown in Recent Runs | 10 most recent |
| Test results per Run Detail page | 100 |
| Final test results stored per run | 20,000 tests / 16 MiB |
| Runs waiting to be saved | 8 runs / 32 MiB in total |
| Saving pending history at shutdown | Up to 5 seconds |
If the save queue is full, a run may not be saved; the dashboard reports that the run was not confirmed saved. The live result is unaffected.
Reports and artifacts
| Item | Limit |
|---|
| Reporter destinations per run | 64 |
| Artifact references per run | 20,000 |
| References per page | 50 |
| PNG preview | 16 MiB; 32,768 pixels per side; 32 million pixels |
| Download | 1 GiB |
When discovery limits are reached, references are recorded as partial; test results are kept.