test(browser): execute controlled Agent Task on pinned Chrome

[seonghobae]

Partial implementation of #28, stacked on PR #65.

Buyer-visible gap

PR #65 provides the deterministic controlled Agent Task fixture, but it is still only static test infrastructure. The next bounded step is to prove that the exact pinned Chrome for Testing build already used by OriginWeave CI can launch an isolated browser profile, load that fixture, perform real WebDriver input/click operations, observe the exact same-document post-condition, and tear the task profile down without inventing a product adapter or model claim.

Dependency

This Draft is stacked on unchanged exact PR #65 head 0888fe3a6ef6da547a37fd075733cc73dc52b2ab and is not independently integrable before that prerequisite.

TDD / RCA sequence

1. Exact test-only head 197ce14a5e407d61ac35b38b45c0cd042dd6278c introduced the Agent Task pinned-browser contract before the executable runner boundary existed.
2. CI run 31453444370, Rust contracts job 93662306248, checked out that exact head and failed in Python repository contracts with 12 intended assertions. The failures were precisely the absent Agent Task fixture path, extension-disabled launch configuration, WebDriver value/click/post-condition commands, cleanup/evidence labels, and the two new runtime functions/constants. This is the observed RED boundary, not a synthetic or predecessor failure.
3. Current exact implementation head f9917cdd8050c9fdf0aefa669f4d981af85479d6 reuses the existing pinned Chrome for Testing 150.0.7871.129 / revision r1639810 runner without modifying workflow authority. It adds a separate Agent Task execution boundary that uses a fresh temporary profile, disables extensions, clears/types the synthetic field through real WebDriver, clicks the submit control, verifies data-state="submitted" and the synthetic result text, emits bounded credential-free repeatability evidence, and proves profile removal after each trial.

Exact-head proof

On unchanged exact head f9917cdd8050c9fdf0aefa669f4d981af85479d6 against unchanged prerequisite #65 head 0888fe3a6ef6da547a37fd075733cc73dc52b2ab:

• CI run 31453647157: success;
• Rust contracts job 93662920741: Python repository contracts, canonical formatting, workspace/all-target check, full tests, strict Clippy and rustdoc all success;
• Production coverage job 93662920759: exact owned production function/line/region/branch enforcement success;
• pinned real-browser run 31453647201, job 93662920705: success using Chrome for Testing 150.0.7871.129 / revision r1639810;
• bounded evidence artifact mv3-compatibility-f9917cdd8050c9fdf0aefa669f4d981af85479d6, digest sha256:ab1ebc01a45cfbc92992bc178f86f7a733c0ce4d8e934850e66929afe18cb8a5: Agent Task 3/3, pass rate 1.0; every trial reports post_condition=true, input_echo_verified=true, extensions_disabled=true, profile_cleaned=true; existing MV3 compatibility also remains 3/3 with all reported surfaces true;
• CodeRabbit exact-head commit status: success;
• no inline review threads are currently returned; and
• GitHub reports this Draft mergeable against its unchanged exact prerequisite base.

No predecessor-head check or review is transferred. This Draft remains stacked and must not be marked independently ready or merged while #65 is active.

Truth boundary

This lane is executable CI evidence, not the final OriginWeave browser runtime. CSS selectors are test-harness locators for the controlled fixture and are not represented as semantic role/name authority. This branch does not implement or claim WebDriver BiDi/CDP product-adapter semantics, OriginWeave semantic observation/node-handle composition, deterministic Agent policy dispatch, trusted telemetry/process attribution, secret brokerage, model execution, persistence, or completion of issue #28. Those remain separate active prerequisites and the later real runtime composition boundary.

[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: d69113eb-382a-4ef7-9f02-1be4e83d3f29

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.}