Diagnose a live webpage with DeepSeek Harness

Prerequisite:
Use only the native Lexmount tools already loaded in the current DSH process for webpage navigation and interaction. Use DSH's file and Git tools only for read-only checks of whether the workspace relates to the target source.

Call lexmount_doctor before any browser action. If it is unavailable, stop with SETUP_REQUIRED and tell the user to complete the Lexmount Browser Bundle setup for this template, restart DSH Web, and resend this prompt. Do not install the Bundle in the current process or switch browsers. If credentials are missing, call lexmount_auth_login, wait for browser authorization, and run doctor again. Continue only when doctor reports ready.

Compatibility:
The current DeepSeek model does not accept image input. Do not call screenshot tools or any tool that returns image content, and do not generate image files. Use only lexmount_browser_snapshot text or DOM results, lexmount_browser_evaluate values, Session or Replay metadata, and text or JSON files as evidence. Replay is for human review only; the Agent must not inspect frames or attach Replay content to the conversation.

Task:
Measure the UI reveal latency on Selenium's official dynamic test page:
https://www.selenium.dev/selenium/web/dynamic.html
After "Reveal a new input" is activated, #revealed must become visible within 500 milliseconds. Set the result to REPRODUCED when elapsed_ms is finite and greater than 500, and NOT_REPRODUCED when it is finite and less than or equal to 500. Compare the deployed behavior with this official source:
https://raw.githubusercontent.com/SeleniumHQ/selenium/trunk/common/src/web/dynamic.html

Execution:
1. Create artifacts/dsh-dynamic-ui-triage. If it already exists, append a UTC timestamp.
2. Use lexmount_session_create to create exactly one temporary normal-mode Lexmount Cloud Browser Session.
3. Open the test page and use lexmount_browser_snapshot to verify that #reveal exists and #revealed is initially hidden. Save the URL, access time, selector presence, visibility, and relevant DOM text to evidence/01-before.json.
4. Do not use a fixed sleep. Before clicking, use lexmount_browser_evaluate to install a read-only timing probe based on performance.now() and MutationObserver. Click #reveal, read elapsed_ms with a bounded evaluate call, and save the start, end, elapsed_ms, sla_ms, and violated values to evidence/02-timing.json.
5. Use another snapshot to verify that #revealed is visible and save the result to evidence/03-after.json.
6. Extract the smallest deployed script fragment containing reveal() and setTimeout to evidence/04-deployed-source.txt. Open the official source, locate the corresponding function, and save it to evidence/05-reference-source.txt. Keep deployed facts separate from reference-source facts, and never assume that public trunk is the deployed commit.
7. Inspect the current workspace Git remote read-only. Record workspace_code_facts only when the workspace clearly belongs to SeleniumHQ/selenium and the target file exists; otherwise mark the relation as unrelated or unknown and stop scanning unrelated code.
8. Write report.md and report.json, separating browser_facts, deployed_source_facts, reference_source_facts, workspace_code_facts, hypotheses, recommended_next_steps, replay, and cleanup.
9. Close the Session and retain its Replay.

Acceptance:
Doctor must be ready; exactly one Session must be created; no image-producing tool may be called; elapsed_ms must be finite; all five evidence files and both reports must exist and be non-empty; all JSON must parse; every fact must have an evidence category; the Session must be closed; and Replay must be retained. If the main task succeeds but cleanup fails, return PARTIAL_SUCCESS. If a valid measurement cannot be obtained, report failure without inventing a result. Never describe the test page's intentional delay as an unknown production incident.

Cleanup:
On success, failure, timeout, or cancellation, close only the Session created by this task. Do not close pre-existing Sessions, delete Replay, modify source code, submit forms, publish content, delete remote data, or perform irreversible actions.

Return:
Report only the status, result, elapsed_ms, SLA threshold, first failed step, report directory, text evidence file count, Replay retention, cleanup result, and sanitized error summary. Do not expose API keys, Project IDs, Session IDs, Replay IDs, connect URLs, WebSocket URLs, OAuth details, or credentials.