Commit Graph

102 Commits

Author SHA1 Message Date
Gregor Žunič 3e6cf7d2ac Refresh browser-use-core during CI sync 2026-06-08 10:54:57 -07:00
Magnus Müller 3c7cdcc385 Remove frozen 2025-10-25 10:44:06 -07:00
Magnus Müller 863aaff487 Less retry 2025-10-25 10:42:13 -07:00
Magnus Müller d6a0193d96 Api keys 2025-10-25 10:38:05 -07:00
Magnus Müller ad6dac41c8 Cache uv 2025-10-25 10:33:47 -07:00
Magnus Müller ecf1b9950b Linter 2025-10-25 10:31:31 -07:00
Magnus Müller 16d8833048 Api key rename 2025-10-25 10:24:40 -07:00
Magnus Müller df4babf9a1 Env keys 2025-10-25 10:21:59 -07:00
Magnus Müller 6e31a05acd Split model tests 2025-10-25 10:16:46 -07:00
Magnus Müller aef11dc6f1 Parallel tests 2025-10-25 09:56:33 -07:00
Magnus Müller 4db51f9e24 Ci cd test update 2025-10-25 09:12:15 -07:00
Magnus Müller 9d2e379af7 Api key 2025-10-24 02:40:35 -07:00
Magnus Müller c1abe88048 Higher timeout + weekly chrome cache 2025-10-24 02:30:29 -07:00
Magnus Müller 9f711b4281 Reduce timeout 2025-10-24 01:41:22 -07:00
Magnus Müller 46e727f7bd Enhance CI workflow by adding conditional caching for Chromium installation
This update introduces a caching mechanism for Chromium binaries in the GitHub Actions workflow. The installation of Chromium will now only occur if it is not already cached, reducing unnecessary downloads and speeding up the CI process. This change aims to optimize the workflow efficiency, particularly during parallel test runs.
2025-10-24 01:28:39 -07:00
Magnus Müller e6cb8e7587 Add explicit uv package caching to speed up CI
The built-in astral-sh/setup-uv cache wasn't working properly,
causing 2m+ downloads of Python packages on every test run.

Added explicit ~/.cache/uv caching keyed on uv.lock hash.

Before: 2m 9s downloading packages (numpy, oci, imageio-ffmpeg, etc)
After: ~5s restoring from cache

Saves ~2 minutes per test job × 40 parallel jobs = 80 runner-minutes saved
2025-10-24 01:25:03 -07:00
Magnus Müller 2c7d3372e1 Add setup-chromium job to pre-cache chromium before parallel test runs
Prevents 40 parallel runners from racing to install chromium simultaneously on cache miss.

Before: 40 runners × 2min = 80 runner-minutes wasted on first run
After: 1 runner installs (2min), then 40 runners use cached version (10s each)

Savings on cache miss: ~78 runner-minutes per workflow run
2025-10-24 01:19:06 -07:00
Magnus Müller a2b0f084be Refactor GitHub workflows to remove Playwright version detection and update cache keys for Chromium binaries based on uv.lock file. 2025-10-24 01:07:39 -07:00
Magnus Müller 2ca0ba38ec Bug playwright not found 2025-10-24 01:03:36 -07:00
Magnus Müller 633fb53b1a Enhance GitHub workflows by adding Playwright version detection and updating cache keys for Chromium binaries based on the installed Playwright version. 2025-10-24 01:00:26 -07:00
Magnus Müller aba7e2cbae cancel tests on new commit 2025-09-08 08:52:25 -07:00
Magnus Müller 6bba023d38 Reduce timeout for GitHub Actions tests from 15 to 10 minutes to prevent hanging. 2025-09-02 00:24:09 -07:00
Magnus Müller 172951b8cd Fix GitHub Actions workflow hanging: add timeouts and force fresh checkout
- Add timeout-minutes to prevent jobs from hanging indefinitely
- Force fresh checkout with fetch-depth: 1 to avoid cache issues
- Add file existence check to handle renamed/deleted tests gracefully
- Add debug output to track test discovery process
2025-09-02 00:20:54 -07:00
Magnus Müller dd6a187fc0 Type 2025-09-02 00:14:20 -07:00
Magnus Müller 616c81e435 Skip none existing tests 2025-09-02 00:14:06 -07:00
Magnus Müller 723c68c20c Test fails if 0% 2025-08-29 18:38:24 -07:00
Magnus Müller 657ea42efd Remove Playwright/Patchright dependencies - use uvx playwright install for browser only
- Updated test.yaml to use uvx playwright install chromium --with-deps --no-shell
- Removed Chrome stable installation and patchright dependencies
- Updated Dockerfile to use uvx playwright temporarily for browser installation
- Updated chromium Dockerfile to use uvx playwright without permanent dependency
- Simplified CI pipeline while maintaining browser functionality
2025-08-26 17:27:41 -07:00
Magnus Müller caa0e7ef1b Rename controller to tools instances 2025-08-26 11:30:39 -07:00
Nick Sweeting f5925b1080 fix browser-use extensions dir to use .config/browseruse like everything else and cache in actions 2025-07-29 13:04:44 -07:00
Nick Sweeting fd07360f57 simpler page title fetching 2025-07-10 18:35:24 -07:00
Nick Sweeting cdca6339f6 try all browsers for evals 2025-07-10 16:24:47 -07:00
Nick Sweeting 4b4b93f6cc tweak chrome used for test.yaml evaluate_Tasks 2025-07-09 15:32:10 -07:00
Nick Sweeting 3bd76cea98 tweak chrome used for test.yaml evaluate_Tasks 2025-07-09 15:26:07 -07:00
Nick Sweeting 435426fc9a bump cache action version 2025-07-08 18:40:36 -07:00
Nick Sweeting b4a8776fec speed up chrome install in CI 2025-07-08 18:19:08 -07:00
Nick Sweeting 1fa7fee4f6 fix cache key for tests 2025-07-08 18:10:14 -07:00
Nick Sweeting 7cf6a26664 fix flipped order 2025-07-08 18:07:43 -07:00
Nick Sweeting 1c6b510f07 use sudo for curl to update 2025-07-08 18:05:20 -07:00
Nick Sweeting 28f0d4d401 use runner arch in cache key 2025-07-08 18:03:11 -07:00
Nick Sweeting b206db41a1 use consistent bin name 2025-07-08 18:01:24 -07:00
Nick Sweeting 32e5430b62 only cache actual binary 2025-07-08 18:00:55 -07:00
Nick Sweeting 14030006db fix missing sudo 2025-07-08 17:57:18 -07:00
Nick Sweeting 4599f815f2 try to cache chrome apt package 2025-07-08 17:55:44 -07:00
Nick Sweeting 3f84d1c460 set in_docker in evals 2025-07-08 06:09:30 -07:00
Nick Sweeting 4d8bdb3dbf install all browser versions for evals and tests 2025-07-08 06:05:20 -07:00
Nick Sweeting 7403c33be3 fix user-data-dir matching 2025-07-08 05:05:49 -07:00
Magnus Müller 8ed1f6cb88 Update failing test 2025-07-08 13:43:12 +02:00
Nick Sweeting fe3af9479a run only one proc for tests for now 2025-06-27 03:02:03 -07:00
Nick Sweeting 74d02c07a7 increase screenshot timeout to default timeout 2025-06-26 02:59:57 -07:00
Nick Sweeting 170c3e0bb7 remove duplicate timeout config 2025-06-26 01:24:03 -07:00