test(browser): locate controlled Agent Task targets semantically

[seonghobae]

Partial implementation of #28, stacked on exact PR #73 head 6ba2d03345fa3153230a7d365fba8284696541a1.

Buyer-visible gap

The controlled pinned-Chromium task verified browser-computed role/name before each action, but it still located the fixture targets by CSS identifiers first. Issue #28 requires the first real workflow to locate the form field by semantic role/name, so the test harness now proves bounded semantic discovery rather than merely validating semantics after selector-based discovery.

Dependency

This Draft targets PR #73 branch test/agent-task-process-set-evidence. Keep Draft while #73/#72/#71/#70/#65 remain active. No prerequisite check, review, status, mergeability, or real-browser result transfers to this branch.

TDD / implementation

1. Exact test-only head e0e4cef2546c0564ba86b6301a3375656ed988ae required a bounded _find_element_by_accessible_role_name helper, a finite candidate limit, exact browser-computed role/name matching, explicit zero/multiple-match failures, and removal of direct #task-text / #submit-task discovery. CI run 31552226676 reached the intended Python repository-contract RED because the helper and semantic discovery path did not exist.
2. Production head 0a10234d32603175344dd337f41bbddb5515beca added only the bounded controlled-fixture locator: enumerate at most 128 WebDriver candidates, validate each W3C element identifier, read computed role/name, and require exactly one exact match before clear/type or click. CI 31553300188 and Manifest V3 Compatibility 31553300076 succeeded on that head.
3. Current exact head 13f49b7fc4f11d0fd851f51d816dc0cc94003b91 adds focused behavior regressions for the helper itself: exact-match success plus zero-match, duplicate-match, oversized candidate-set, and malformed-candidate fail-closed cases.

The helper uses the existing W3C WebDriver HTTP element enumeration only as a bounded fixture transport, then filters by browser-computed role/name. This does not become a product selector API. A later product adapter must use a versioned browser protocol boundary rather than treating this compatibility-runner helper as durable authority.

Exact-head proof

On unchanged exact head 13f49b7fc4f11d0fd851f51d816dc0cc94003b91 against unchanged exact prerequisite #73 head 6ba2d03345fa3153230a7d365fba8284696541a1:

• CI run 31553583901: success, including repository contracts, canonical formatting, locked workspace/all-target check, full tests, strict Clippy, rustdoc, and exact owned production function/line/region/branch coverage;
• Manifest V3 Compatibility run 31553583902: success against the pinned real-Chromium lane;
• GitHub reports the Draft mergeable against its exact prerequisite base; and
• no inline review threads are currently returned.

No predecessor-head or prerequisite result is transferred.

Truth boundary

This slice is executable pinned-browser evidence only. It does not implement the production WebDriver BiDi adapter, expose CSS/XPath/model-generated selectors as OriginWeave authority, mint node handles, authorize or dispatch typed actions, prove source/network provenance, or complete issue #28.

[coderabbitai]

Important

Review skipped

Draft detected.

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

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: 6b87e1ae-7de6-457e-8aad-49d1d9aee773

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

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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