[Aaryan430]

Problem

Asked to "scan bstackdemo.com on WCAG 2.2 AAA with best practices on", the agent couldn't. startAccessibilityScan only exposed advancedRules, so every scan ran at the API's defaults: WCAG 2.1 AA, best practices off, needs review on.

Change

startAccessibilityScan takes three new optional arguments, passed to the Website Scanner create-scan API under scanSettings next to advancedRules:

Argument	Values	API default when left out
wcagVersion	wcag2a … wcag22aaa, rgaa41 (the backend's WCAG_VERSION_LABEL_MAP)	wcag21aa
bestPractices	true / false	false
needsReview	true / false	true
advancedRules (existing)	true / false	true

• Settings sent: only the ones given; the rest take the API's own defaults.
• The answer: it now has a line with the settings read back from the run, e.g. Scan settings, as the scan ran: WCAG 2.2 AAA, best practices on, needs review on, advanced rules off. The agent can then report the settings instead of guessing them. If the run's settings can't be read, the line names only what was asked for.
• Not included, experimental rules: the public create-scan API has no field for them. They're a user/team-level Website Scanner setting (disableA11yEngineExperimentalRules) that only the dashboard can change. That needs a backend change first.

Verification

• New tests/tools/accessibilityScanSettings.test.ts:
  • the scanner sends the settings it is given, and only those
  • the tool passes them through and states them
  • wcagVersion accepts only the API's values
• Argument descriptions stay within the 120-character budget.
• npm run build passes on Node 22: 1358 tests.
• Live: startAccessibilityScan on bstackdemo.com with wcagVersion: "wcag22aaa" and bestPractices: true. The API's run summary shows wcagVersion: "wcag22aaa", bestPractices: true, needsReview: true, advanced: true. That's 91 issues, against 81 at the defaults.

Changes after review (Shreyas)

• The settings line reads the run's settings back from its summary rather than labelling defaults from a table here. This also shows advanced rules that a plan turned off. Checked live on three runs.
• A 422 passes on the API's own reason. "A scan with this name already exists" is used only for a 422 that gives none.
• main merged in. npm run build passes: 1360 tests.

🤖 Generated with Claude Code

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

• Source mode: Auto
• Repository: Unavailable
• Branch: Unavailable
• File: No configuration file from this repository applied
• Commit: Unavailable
• Layers: Central YAML (base), Organization UI (inherited), Workspace UI (inherited); Enforced overrides: Organization
• Review profile: ASSERTIVE
• Plan: Enterprise
• Run ID: c4315199-784b-4a4f-b771-7d3945871a83

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

_{Comment @coderabbitai help to get the list of available commands.}

[sarve-shreyas]

Review: #513

The schema change is additive, so existing callers are unaffected. CI shows only CodeQL and CodeRabbit (auto-review skipped), so the test counts in the description aren't CI-verified.

Medium

• scanner.ts catch block: any HTTP 422 is reported as "A scan with this name already exists…". With the new settings, a wcagVersion the backend rejects (e.g. rgaa41 on a plan without it) would send the agent into a rename loop. Check err.response.data and otherwise surface the API message. (Whether the API returns 422 for bad settings is unverified.)

Low-Medium

• scanSettingsNote / SCAN_DEFAULTS are hard-coded, so the "(default)" labels report what was requested, not what ran. Account- or team-level settings could differ. Read the settings from the create-scan/run-summary response, or reword to "API default (assumed)".

Low

• Only wcag22aaa was exercised live. rgaa41, wcag2a etc. are unverified. ScanSettings.wcagVersion is typed string rather than keyof typeof WCAG_VERSIONS.
• Missing tests: a 4xx/422 with settings, and bestPractices: false. The local-URL branch looks correct by reading but is untested.
• The advancedRules description text was reworded, which changes the schema text clients see.

AI-assisted read-only review of the diff. Items marked (unverified) were not checked against the live APIs.

[Aaryan430]

Thanks @sarve-shreyas. Both addressed in 79cc46d:

• Medium (422): a 422 now passes on the API's own reason (Failed to start scan: <message>). The "a scan with this name already exists" text is used only for a 422 that has no reason. Both cases are tested.
• Low-Medium ((default) labels): removed. After the scan finishes, the settings are read back from the run summary, and the note reports them as the scan ran: Scan settings, as the scan ran: WCAG 2.2 AAA, best practices on, needs review on, advanced rules off. This also shows advanced rules that a plan turned off. If the summary can't be read, the note names only what was asked for. I checked the read-back live against three recent runs, rgaa41 and wcag21aa, and confirmed that an unknown run returns nothing.

origin/main merged in. Build passes: lint, prettier, 1360 tests, tsc.

[Aaryan430]

This has been merged into #502 (c7e454d), which now ships it to main together with #509, #510 and #492. It conflicted with #509 in accessibility.ts, where both added new helpers, and both were kept. The scan-settings line is also given when a scan's export is still being generated. The combined build passes (1400 tests).