docs: instruct agents to save test output instead of re-running tests

Adds guidance to always pipe expensive test commands (e2e, web, feature)
through tee to a log file, so agents can re-read the output for analysis
instead of re-running the full test suite with different grep/tail filters.
This commit is contained in:
kolaente
2026-02-25 09:21:01 +01:00
parent 91e68168d0
commit c8ea673653

View File

@@ -207,6 +207,19 @@ mage test:e2e "--grep menu" # filter by name
mage test:e2e "--headed tests/e2e/misc/menu.spec.ts" # headed mode
```
**IMPORTANT: Always save test output to a file.** E2E tests are expensive (they rebuild the API, start servers, run browsers, etc.). NEVER re-run tests just to look at the output differently (e.g., with different `grep`/`tail` filters). Instead, save the output on the first run and then read the file:
```bash
# First run: save output to a file
mage test:e2e "tests/e2e/misc/menu.spec.ts" 2>&1 | tee /tmp/e2e-output.log
# Subsequent analysis: read the file, don't re-run
cat /tmp/e2e-output.log | grep -E '(passed|failed)'
cat /tmp/e2e-output.log | tail -20
```
This also applies to `mage test:web`, `mage test:feature`, and `mage test:filter`.
Set `VIKUNJA_E2E_SKIP_BUILD=true` to skip rebuilding the API binary when iterating on frontend-only changes.
## Swagger API Documentation