Dream Loop skill
Iterate a running app toward a concrete target screenshot.
Community · Anshu Chimala| Author | Anshu Chimala |
|---|---|
| Licence | MIT |
| Source | achimala/dream-loop · SKILL.md on GitHub |
| Works in | Claude Code · Codex, Gemini CLI, Cursor, Copilot |
| Read from the source on |
dream-loop: Iterate a running app toward a concrete target screenshot.
It selects one workflow and stores a target image before repeated build-and-compare rounds.
From the skill's SKILL.md: # dream-loop
What it does
- Lock a visual target, implement and compare screenshots.
What you get
- App changes, screenshots and progress context.
When to use it
- Pursue a specific visual result.
When not to use it
- Claiming pixel-perfect completion without comparison.
What it needs
- Project/build tools, screenshot access and either a supplied target image or image-generation access.
Files in the skill
| File or folder | What it is for |
|---|---|
SKILL.md | Entry-point instructions and supporting workflow. |
Licence as found in the repository on 2026-11-16: MIT.
Example prompts
Type one of these after installing; naming the skill makes sure it is used.
- Match this target screenshot.
- Improve this scene toward the reference.
- Compare the next iteration with target.png.
How to install dream-loop
Prepare the workflow
Copy root references. Generate only when authorized and necessary; a supplied target needs no generation. Do not load incompatible workflow modes together.
A skill is a folder with a SKILL.md file. You install it once; the agent reads its short description at the start of each session and loads the full instructions only when a request matches. Pick the place you use:
Claude Code: copy the folder into your skills folder
Put the skill's folder in ~/.claude/skills/ to use it in every project on this computer, or in .claude/skills/ inside a repository to share it with everyone who works there (commit it).
git clone --depth 1 https://github.com/achimala/dream-loop.git
mkdir -p ~/.claude/skills/dream-loop
git -C dream-loop archive HEAD | tar -x -C ~/.claude/skills/dream-loopClaude Code notices new or changed skills in these folders during the session. If the skills folder did not exist when the session started, run /reload-skills once. Then type /dream-loop or just describe the task.
On Windows, run the commands in Git Bash or WSL, or copy the folder into the same place with File Explorer.
Codex, Gemini CLI, Cursor and GitHub Copilot
These agents read the same SKILL.md folders. The one folder all four accept inside a project is .agents/skills/, so this copies the skill there:
git clone --depth 1 https://github.com/achimala/dream-loop.git
mkdir -p .agents/skills/dream-loop
git -C dream-loop archive HEAD | tar -x -C .agents/skills/dream-loop| Agent | Project folder | Your user folder | Notes |
|---|---|---|---|
| OpenAI Codex | .agents/skills/ | ~/.agents/skills/ | Type $ or run /skills to pick a skill; Codex also chooses one when your task matches. |
| Gemini CLI | .agents/skills/.gemini/skills/ | ~/.agents/skills/~/.gemini/skills/ | Or install from Git with gemini skills install <git URL> --path <folder> --consent. Google replaced Gemini CLI with Antigravity CLI for free-tier and Google One users on 18 June 2026; Antigravity uses .agents/skills/ in a project and ~/.gemini/config/skills/ for you. |
| Cursor | .agents/skills/.cursor/skills/ | ~/.agents/skills/~/.cursor/skills/ | Type / in Agent chat to pick a skill. Cursor also reads Claude Code's .claude/skills/ folders. |
| GitHub Copilot | .github/skills/.agents/skills/.claude/skills/ | ~/.copilot/skills/~/.agents/skills/ | Works in Copilot CLI, the cloud agent, and agent mode in VS Code, Visual Studio and JetBrains (IDE support depends on the version). Or run gh skill install OWNER/REPOSITORY SKILL (public preview, GitHub CLI 2.90.0 or later). |
Sources for these steps
Checked against each vendor's documentation on 2026-11-14:
Use a concrete target
For existing products, refine a capture of the current app instead of inventing an unrelated visual direction.
Compare between rounds
Keep context and target stable. Screenshot evidence should determine which differences remain.
Questions about dream-loop
Does copying the skill install its tools?
No. Install and configure external requirements separately.
Is full cloud execution verified here?
No. The page offers source-based local-agent guidance, not a tested cloud runtime.
Must it generate a target?
No. A supplied target image can be used directly.
Does it always finish within a budget?
No. Report meaningful progress and remaining differences honestly.
More AI skills
Related skills
Modern Web Design skill
Community · freshtechbroA 2024–2025 reference for responsive layout, purposeful motion, accessibility and web design patterns.Awwwards-Style 3D skill
Community · awwwards-3d contributorsBuild cinematic scroll-driven 3D websites with a pinned stack.Design Auditor skill
Community · Ashutos1997Review interface code and Figma designs with evidence and actionable fixes.Three.js Fundamentals skill
Community · CloudAI-XScenes, cameras, renderer setup, transforms and cleanup.Three.js Geometry skill
Community · CloudAI-XBuilt-in shapes, BufferGeometry, instancing, batching and point clouds.Three.js Materials skill
Community · CloudAI-XUnlit, shaded, PBR, transparent and node-based materials.