1 00:00:00,000 --> 00:00:03,336 This guide covers Gatekeeper Q A from the 2 00:00:03,336 --> 00:00:06,672 command line and in GitHub Actions. We will 3 00:00:06,672 --> 00:00:07,506 configure an 4 00:00:07,506 --> 00:00:10,842 explicit page, inspect a failing result, fix a 5 00:00:10,842 --> 00:00:14,178 button label, and retrieve a successful retest. Browser 6 00:00:14,178 --> 00:00:15,012 captures are 7 00:00:15,012 --> 00:00:18,792 from real runs. Command cards are readable instructions; 8 00:00:18,792 --> 00:00:22,573 local result cards summarize retained execution logs. Narration 9 00:00:22,573 --> 00:00:23,518 is synthetic. 10 00:00:23,518 --> 00:00:26,054 Use Node J S twenty four with N 11 00:00:26,054 --> 00:00:28,591 P M. In a new project directory, initialize 12 00:00:28,591 --> 00:00:29,225 package dot 13 00:00:29,225 --> 00:00:31,762 J S O N, install Gatekeeper Q A 14 00:00:31,762 --> 00:00:34,299 C L I version zero point three point 15 00:00:34,299 --> 00:00:34,933 zero, then 16 00:00:34,933 --> 00:00:37,470 install Playwright Chromium. In an existing project, skip 17 00:00:37,470 --> 00:00:40,007 N P M init. The clean GitHub runner 18 00:00:40,007 --> 00:00:40,641 installation shown 19 00:00:40,641 --> 00:00:47,349 later also verifies this published package. 20 00:00:47,349 --> 00:00:49,845 Start your application using its normal development command, 21 00:00:49,845 --> 00:00:52,341 and wait until its page is available. Save 22 00:00:52,341 --> 00:00:52,965 gatekeeper Q 23 00:00:52,965 --> 00:00:55,461 A dot J S O N with the 24 00:00:55,461 --> 00:00:57,957 actual U R L. This example uses localhost 25 00:00:57,957 --> 00:00:58,581 port three 26 00:00:58,581 --> 00:01:01,077 thousand. Replace that address if your app uses 27 00:01:01,077 --> 00:01:03,574 another port. The C L I checks explicit 28 00:01:03,574 --> 00:01:04,198 U R 29 00:01:04,198 --> 00:01:10,814 Ls, not a whole-site crawl. 30 00:01:10,814 --> 00:01:13,243 Run the C L I with your configuration 31 00:01:13,243 --> 00:01:15,673 and a separate before output directory. Open report 32 00:01:15,673 --> 00:01:16,280 dot H 33 00:01:16,280 --> 00:01:18,709 T M L after the command finishes. In 34 00:01:18,709 --> 00:01:21,139 our local demonstration, the unnamed button produced a 35 00:01:21,139 --> 00:01:21,746 critical button-name 36 00:01:21,746 --> 00:01:24,175 finding and exit code one. The retained transcript 37 00:01:24,175 --> 00:01:26,605 records that real result; the equivalent N P 38 00:01:26,605 --> 00:01:27,212 X command 39 00:01:27,212 --> 00:01:33,678 is shown here. 40 00:01:33,678 --> 00:01:36,878 Exit zero does not mean zero accessibility issues. 41 00:01:36,878 --> 00:01:41,448 In GatekeeperQA CLI version 0.3.0, 42 00:01:41,448 --> 00:01:44,108 exit zero means pass or warning. 43 00:01:44,108 --> 00:01:46,968 Exit one means the report status is fail. 44 00:01:46,968 --> 00:01:50,428 Exit two means a configuration or runtime error. 45 00:01:50,428 --> 00:01:51,768 Always read the report. 46 00:01:51,768 --> 00:01:56,125 Automated checks do not replace human testing. 47 00:01:56,125 --> 00:01:58,985 Give the button meaningful visible text and run 48 00:01:58,985 --> 00:02:01,845 the same scope into an after directory. Our 49 00:02:01,845 --> 00:02:02,560 local retest 50 00:02:02,560 --> 00:02:05,420 returned pass and exit zero. Preserve the baseline. 51 00:02:05,420 --> 00:02:08,280 Reproduce the affected page state and test the 52 00:02:08,280 --> 00:02:08,995 interaction manually, 53 00:02:08,995 --> 00:02:12,712 including a screen reader. A passing automated result 54 00:02:12,712 --> 00:02:16,429 does not prove the entire experience is accessible. 55 00:02:16,429 --> 00:02:19,085 Copy the downloadable accessibility workflow into dot github 56 00:02:19,085 --> 00:02:21,741 slash workflows slash accessibility dot Y A M 57 00:02:21,741 --> 00:02:22,405 L in 58 00:02:22,405 --> 00:02:25,061 your repository. Commit it to the default branch 59 00:02:25,061 --> 00:02:27,716 so Run workflow is available. The template uses 60 00:02:27,716 --> 00:02:28,380 a manual 61 00:02:28,380 --> 00:02:31,036 trigger, read-only repository permission, Node twenty four, the 62 00:02:31,036 --> 00:02:33,692 pinned C L I, and seven-day report artifacts. 63 00:02:33,692 --> 00:02:34,356 It does 64 00:02:34,356 --> 00:02:41,331 not deploy your application. 65 00:02:41,331 --> 00:02:44,065 Open Actions, select the workflow, and choose Run 66 00:02:44,065 --> 00:02:46,799 workflow. Select the branch containing your intended workflow 67 00:02:46,799 --> 00:02:47,483 version and 68 00:02:47,483 --> 00:02:50,217 enter an authorized public page. Our recording uses 69 00:02:50,217 --> 00:02:52,951 an existing verification workflow on the review branch, 70 00:02:52,951 --> 00:02:53,635 with C 71 00:02:53,635 --> 00:02:56,369 L I zero point three point zero. The 72 00:02:56,369 --> 00:02:59,103 downloadable template has the same scan and artifact 73 00:02:59,103 --> 00:02:59,787 steps, with 74 00:02:59,787 --> 00:03:06,939 a manual trigger. 75 00:03:06,939 --> 00:03:09,203 The first GitHub run failed on the original 76 00:03:09,203 --> 00:03:11,467 demonstration page. Expand Scan authorized U R L 77 00:03:11,467 --> 00:03:12,033 and Enforce 78 00:03:12,033 --> 00:03:14,297 result to see the result and exit code. 79 00:03:14,297 --> 00:03:16,561 Save reports runs before the gate is enforced, 80 00:03:16,561 --> 00:03:17,127 so a 81 00:03:17,127 --> 00:03:19,391 failed accessibility result still leaves evidence. Open the 82 00:03:19,391 --> 00:03:21,654 artifact on the run summary rather than relying 83 00:03:21,654 --> 00:03:22,220 on the 84 00:03:22,220 --> 00:03:28,314 red status alone. 85 00:03:28,314 --> 00:03:30,862 After publishing the button fix, we ran the 86 00:03:30,862 --> 00:03:33,411 same workflow and public U R L again. 87 00:03:33,411 --> 00:03:34,048 This run 88 00:03:34,048 --> 00:03:36,596 succeeded, with one report artifact. Open the accessibility 89 00:03:36,596 --> 00:03:39,145 job for its scan output, then download accessibility 90 00:03:39,145 --> 00:03:39,782 reports from 91 00:03:39,782 --> 00:03:42,330 the summary. Extract the archive and open report 92 00:03:42,330 --> 00:03:44,879 dot H T M L; retain the machine-readable 93 00:03:44,879 --> 00:03:45,516 files for 94 00:03:45,516 --> 00:03:52,250 your tools. 95 00:03:52,250 --> 00:03:54,459 The saved artifact contains the generated report files. 96 00:03:54,459 --> 00:03:56,668 Download it before its seven-day retention ends. The 97 00:03:56,668 --> 00:03:57,220 runner installed 98 00:03:57,220 --> 00:03:59,428 the public N P M package and Chromium 99 00:03:59,428 --> 00:04:01,637 cleanly; the two recorded runs took fifty four 100 00:04:01,637 --> 00:04:02,189 and forty 101 00:04:02,189 --> 00:04:04,398 one seconds. Check your own GitHub plan for 102 00:04:04,398 --> 00:04:06,607 runner billing. A completed job proves this run, 103 00:04:06,607 --> 00:04:07,159 not every 104 00:04:07,159 --> 00:04:13,129 page or state. 105 00:04:13,129 --> 00:04:15,506 For pull-request testing, first build and start the 106 00:04:15,506 --> 00:04:17,884 changed application on the runner, or deploy a 107 00:04:17,884 --> 00:04:18,478 preview and 108 00:04:18,478 --> 00:04:20,855 wait for it. Point the scan at that 109 00:04:20,855 --> 00:04:23,232 changed version. A manual scan of production is 110 00:04:23,232 --> 00:04:23,826 not a 111 00:04:23,826 --> 00:04:26,203 pull-request check. Add the appropriate trigger and branch 112 00:04:26,203 --> 00:04:28,581 protection deliberately. This guide does not configure a 113 00:04:28,581 --> 00:04:29,175 P R 114 00:04:29,175 --> 00:04:35,523 bot or automatic comments. 115 00:04:35,523 --> 00:04:38,339 You now have a repeatable C L I 116 00:04:38,339 --> 00:04:41,154 and Actions workflow: explicit scope, a clear gate, 117 00:04:41,154 --> 00:04:41,858 report artifacts, 118 00:04:41,858 --> 00:04:44,674 and a separate retest. Use the browser tester 119 00:04:44,674 --> 00:04:47,489 guide to record manual observations and compare saved 120 00:04:47,489 --> 00:04:48,193 reports. Download 121 00:04:48,193 --> 00:04:51,008 the workflow and read the developer quick start 122 00:04:51,008 --> 00:04:53,823 on gatekeeper Q A dot com. Keep human 123 00:04:53,823 --> 00:04:54,527 testing part 124 00:04:54,527 --> 00:05:01,862 of every release.