8000 test(clicks): a deep research journey, and the switch keeps the cursor in the box by adeelahmad · Pull Request #561 · adeelahmad/lense · GitHub
[go: up one dir, main page]

Skip to content

test(clicks): a deep research journey, and the switch keeps the cursor in the box - #561

Merged
adeelahmad merged 5 commits into
mainfrom
claude/project-thread-r960xc-c9-research
Oct 18, 2026
Merged

adeelahmad merged 5 commits into
mainfrom
claude/project-thread-r960xc-c9-research

Conversation

@adeelahmad
Copy link
Copy Markdown
Owner

Requested by Adeel · project thread

Description

Before: the clicks survey had no deep research journey, and turning on the composer's Deep research switch moved the cursor out of the question box, so a person had to click back into it before typing the goal.

After: the switch (on and off) keeps the cursor in the box, and the survey has a chat-research journey: Chat › Deep research › type the goal › Enter › Approve the plan › Save as note. It takes 4 clicks on a computer and 6 taps on a phone, and is in the baselines and in docs/requirements.md (deep research ≤ 4 clicks).

The page survey found navigation links by their accessible name, so the report's note (named "Written by the assistant" and its title) stopped the run. It now finds them by where they go and visits each kind of page once, so the notes the journeys leave in the navigation count as one page, as before.

How: the survey's stand-in model answers structured calls. A deep research planning prompt gets a plan for its goal with no questions to answer first, so the journey can approve it as it is; any other structured call gets {}. The composer's two research buttons focus the box after toggling.

Motivation and Context

Part of the chat-app gaps (C9: click journeys for each chat feature).

Local runs on this commit: the full survey on a computer and a phone (both pass; the only baseline change is the new chat-research, and the pages surveyed stay 75 on each), chat-research.test.tsx (14 passed), ci/check_requirements.py, Prettier and ESLint.

Types of changes

  • Bug fix (non-breaking change which fixes an issue)
  • New feature (non-breaking change which adds functionality)
  • Breaking change (fix or feature that would cause existing functionality to change; ! in the title)

Checklist:

  • The title is a Conventional Commit.
  • My code follows the code style of this project (make hooks checks it on commit).
  • I added or updated tests: the chat-research journey, and a focus check in chat-research.test.tsx.
  • My change requires documentation updates, and I have updated docs/ accordingly: assistant.md (the cursor stays in the box, and a refine-later note) and requirements.md (the deep research row).
  • Users need more than the title to understand it: n/a, the title says it.

Done questions (AGENTS.md)

  • Goal: delivers what the user wanted, merged on green CI. Not merged yet.
  • Agents: n/a, deep research already has POST /chats/{cid}/research and MCP start_research; this is the survey and a focus fix.
  • Zero setup: nothing new to install or configure.
  • Fewest clicks and keystrokes: one click fewer for deep research (no click back into the box), now measured.
  • Configurable: no new settings.
  • Light: nothing new runs; the stand-in model runs only in the survey.
  • Private and secure: n/a, no new surface.
  • Cost: n/a, no new model calls.
  • User decides: the plan still waits for approval; defaults unchanged.
  • Upgrades: n/a, no stored-data change.
  • Honest: the requirement names checks that exist (ci/check_requirements.py).
  • Tests, docs and generated files updated (no API change).
  • Open standards; dependencies keep Lens MIT: no new dependencies.
  • Result reported with evidence; "refine later" items written down: one way to read a report aloud (docs/assistant.md, Deep research).
  • Reuse: the journey uses the survey's own meter and helpers.

Generated by Claude Code

…r in the box

The survey's stand-in model answers a deep research plan, the chat-research
journey plans, approves and saves a report as a note, and the page survey
finds navigation links by where they go (a note the assistant wrote has more
in its name than its title) and visits each kind of page once. The composer's
Deep research switch puts the cursor back in the question box.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HXPWXcSCQb9hWPr7xkZu7B
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HXPWXcSCQb9hWPr7xkZu7B
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants

0