Skip to main content
prompt GP ChatGPT

phase-10d-chat-ui-polish-research — Claude Code prompt

The phase-10d-chat-ui-polish-research prompt guides a review of Phantom’s chat UI using screenshots and project files, then asks for a design direction and an implementation plan.

Apache-2.0 ★ 1,497 updated 2026-06-16

Install into Claude Code · Manually: put the files in place All methods and platforms ↓
.claude/commands/phase-10d-chat-ui-polish-research.md
Open on GitHub

What phase-10d-chat-ui-polish-research does

The prompt asks Claude Code to inspect Phantom’s chat interface, review the screenshots, and trace the relevant UI components. The resulting research document should cover the visual system, tool cards, message composer, mobile layout, and a plan for implementation and verification.

Required sections include a screenshot-based critique, analysis of real product references, a proposed design direction, component-level changes, responsive behavior, regression checks, and a builder prompt seed. Source code must remain untouched; the sole write target is research/chat-experience/phase-10d-chat-ui-polish-research.md.

Claude Code is instructed to read the listed project files and inspect the supplied screenshots first. If temporary images can no longer be opened, it should use the textual UI critique instead. Access to the project files matters: without it, the review cannot ground its recommendations in the current implementation.

What it helps with

  • review chat UI issues against screenshots
  • separate CSS work from data-contract changes
  • define hierarchy for messages, tools, and run status
  • plan component and mobile-layout updates
  • specify accessibility and visual regression checks

Example prompts for the agent

  • Research Phantom’s chat UI with the phase-10d-chat-ui-polish-research prompt and write the report to its specified path.
  • Compare the tool cards with the screenshots and propose changes for Phase 10D.
  • Create a responsive, accessibility, and visual regression test plan for Phantom chat.

How to install phase-10d-chat-ui-polish-research

Manually: put the files in place

.claude/commands/phase-10d-chat-ui-polish-research.md

Questions about phase-10d-chat-ui-polish-research

How do I install phase-10d-chat-ui-polish-research in Claude Code?

Save the prompt as .claude/commands/phase-10d-chat-ui-polish-research.md. Claude Code can then use it as a slash command.

What does phase-10d-chat-ui-polish-research do?

It asks Claude Code to examine Phantom’s chat UI through screenshots and project code, then write research with a design direction and implementation and test plans. It forbids source-code edits.

What does the prompt need to work?

Claude Code needs access to the listed project files and screenshots. If the temporary images have expired, the prompt says to rely on the written description of the UI issues.

Source text

phase-10d-chat-ui-polish-research.md · Apache-2.0
ultrathink. ultrathink. ultrathink.

You are a principal engineer, principal architect, AND principal product
manager at Anthropic, all three at once. You are researching a full Phantom
chat UI polish pass for an AI co-worker with its own computer.

No context anxiety. No token limits. No time pressure. No cost anxiety. No v2
thinking. There is no v2. Build it right. Take as long as you need. This prompt
is your starting point, not your ceiling. Exceed it.

You are fully autonomous. You may inspect code, screenshots, and current UI
structure. You may use web research for best-in-class public product patterns,
but only from real products or primary sources, not SEO blogs. Do not edit
source code. Your only write target is:

`research/chat-experience/phase-10d-chat-ui-polish-research.md`

## Mission

The current chat UI works, but it does not yet feel like a polished, trusted
agent workspace. The screenshots show rough edges:

1. Header/sidebar borders feel heavy and inconsistent.
2. Message bubbles and code blocks are too coarse.
3. Tool cards show that a tool exists, but their expanded content is not
   visually useful enough.
4. The composer feels like a generic chat box, not a work surface.
5. The visual hierarchy around thinking, tools, and final answers needs a
   better system.

Research how to make Phantom chat feel closer to the best AI work products,
while staying quiet, dense, and operational. Think Linear, Notion, Vercel,
Claude Code, OpenAI Codex, Cursor, and high-quality developer consoles.

## User Screenshot References

Use these screenshots as concrete evidence:

1. `/var/folders/d3/1cwh67m14qd_tszmmjlktqx00000gp/T/TemporaryItems/NSIRD_screencaptureui_u7zK8O/Screenshot 2026-04-30 at 3.18.38 PM.png`
2. `/var/folders/d3/1cwh67m14qd_tszmmjlktqx00000gp/T/TemporaryItems/NSIRD_screencaptureui_jOdIT4/Screenshot 2026-04-30 at 3.19.17 PM.png`
3. `/var/folders/d3/1cwh67m14qd_tszmmjlktqx00000gp/T/TemporaryItems/NSIRD_screencaptureui_PYcg1I/Screenshot 2026-04-30 at 3.19.23 PM.png`

If a file path fails because macOS temporary files expired, use the textual
description above and note that the image could not be reopened.

## Required Reading

Read these files in order:

1. `/Users/truffle/.claude/AGENTS.md` around 70 lines.
2. `/Users/truffle/.claude/CLAUDE.md` around 300 lines.
3. `research/chat-experience/phase-10a-synthesis.md` around 351 lines.
4. `chat-ui/src/index.css` or equivalent global styles.
5. `chat-ui/src/components/app-shell.tsx`.
6. `chat-ui/src/components/sidebar.tsx` if present.
7. `chat-ui/src/components/message-list.tsx` around 113 lines.
8. `chat-ui/src/components/message.tsx`.
9. `chat-ui/src/components/assistant-message.tsx`.
10. `chat-ui/src/components/run-activity-row.tsx` around 91 lines.
11. `chat-ui/src/components/tool-call-card.tsx` around 144 lines.
12. `chat-ui/src/components/thinking-block.tsx`.
13. `chat-ui/src/routes/session-route.tsx` around 93 lines.
14. `chat-ui/src/components/chat-composer.tsx` if present.

Use `rg --files chat-ui/src` to locate renamed components.

## Questions To Answer

1. What are the current visual system problems visible in the screenshots?
2. Which problems are pure CSS/layout?
3. Which problems require data model or event improvements?
4. What should the information architecture be for:
   - user messages
   - assistant final answers
   - active run status
   - thinking
   - tool calls
   - artifacts
   - compaction
   - errors
5. Should tool cards live inline, in a timeline rail, in a details drawer, or
   in a hybrid layout?
6. How should expanded tool details look when the content is stdout, command,
   browser activity, web fetch, file edit, or artifact creation?
7. How should mobile differ from desktop?
8. What should be changed in Phase 10D, and what should wait until the data
   contract is richer?
9. What accessibility and keyboard details matter?
10. What visual regression/browser checks are required?

## Acceptance Criteria

Your research file must include:

1. A screenshot-backed critique.
2. A best-in-class reference analysis from real products.
3. A proposed Phantom chat design system direction.
4. A component-level implementation plan.
5. A responsive layout plan.
6. A test and screenshot verification plan.
7. A "builder prompt seed" section.

## Anti-Patterns

Do not:

1. Make a marketing page or decorative hero. This is an operational tool.
2. Use oversized cards, one-note palettes, gradient blobs, or decorative noise.
3. Hide important work state behind tiny icons only.
4. Show raw chain-of-thought or raw secret-bearing logs by default.
5. Let long paths, code, or buttons overflow their containers.
6. Put cards inside cards.
7. Ignore mobile.
8. Redesign around vibes without grounding in current code and screenshots.

ultrathink before writing. Self-review the output against every acceptance
criterion. Your final answer should list the output file you wrote and the top
five findings only.

© ghostwright · Apache-2.0 · ghostwright/phantom

Similar prompts

All prompts for Frontend and UI →

From the same set38

The whole set →
phase-10i-transcript-recovery-builder The phase-10i-transcript-recovery-builder prompt for Claude Code builds transcript recovery for Phantom on Murph, letting agents search chat history after context compaction. GP ChatGPT ★ 1,497repo stars thread Trace how your thinking on a topic changed over time. The skill turns memory entries into a chronological account with marked turning points. CX Codex +1 ★ 1,497repo stars phase-10a-chat-pipeline-research phase-10a-chat-pipeline-research — a slash command that walks Phantom's chat pipeline from SDK events through SSE to React rendering. You get a file-grounded report, and no code is touched. GP ChatGPT ★ 1,497repo stars overheard overheard scans two weeks of stored memory to surface unfulfilled commitments and prepares followup drafts. CX Codex +1 ★ 1,497repo stars ritual Detects recurring behaviors from your session history and transforms them into scheduled tasks. If you regularly do something on a cadence, ritual proposes automating it. CX Codex +1 ★ 1,497repo stars show-my-tools show-my-tools audits the agent environment and lists loaded skills, memory files, plugins, hooks, and subagents. It also outputs a settings summary and dashboard links. CX Codex +1 ★ 1,497repo stars phase-10h-chat-integrity-builder The phase-10h-chat-integrity-builder prompt directs Claude Code to fix transcript integrity in the Phantom chat, including attachment handling and SDK errors. It targets an existing project. GP ChatGPT ★ 1,497repo stars phase-10a-activity-correctness-builder The phase-10a-activity-correctness-builder prompt for Claude Code governs activity visibility during long-running requests, showing users that work continues when the interface is quiet. GP ChatGPT ★ 1,497repo stars

Project sponsors

Companies that keep this analytics open