D2c

Community

BIAsia

d2c

c2d2c d2c pipeline — the standard eight-step workflow for implementing or updating UI from Figma designs. Use EVERY time the user provides figma.com/design/... links asking to build, update, restyle, or restore a UI to match a design. Covers survey → plan review → spec extraction → 1:1 build → regression page → motion review → screenshot acceptance → cleanup + MR. Not for exporting code INTO Figma (that is c2d2c:c2d).

Overview

Publisher	BIAsia
Repository	`c2d2c`
Skill name	`d2c`
Stars	58
Forks	2
Bundled files	1
License	MIT
Links	Skill source Repository Homepage

Markdown instructions
A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.
Works with any LLM
AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.
1 bundled files
Scripts, templates, and references the model can read while it works. Files are read-only and never executed.
Open source
Published by BIAsia on GitHub. Read the source before you install it.

Installation

Install the D2c AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

Open the app and go to Plugins → Skills.
Choose "Install from GitHub".
Paste the skill folder URL below and confirm.
https://github.com/BIAsia/c2d2c/tree/main/skills/d2c
Enable the skill in any chat where you want it available.

Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills

git clone --depth 1 https://github.com/BIAsia/c2d2c.git /tmp/c2d2c
mkdir -p .claude/skills
cp -r /tmp/c2d2c/skills/d2c .claude/skills/d2c

Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable D2c in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads D2c on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so D2c is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

Loaded only when it is needed
The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.
Switch models mid-chat
Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

c2d2c:d2c — design to code, 1:1

Project parameters

Resolve parameters from C2D2C.md at the repo root (fallback: .claude/c2d2c.md). If neither exists, offer to bootstrap one from the template at ../../templates/C2D2C.template.md (plugin root; standalone installs without that file fetch https://raw.githubusercontent.com/BIAsia/c2d2c/main/templates/C2D2C.template.md): fill what the repo itself reveals (token files, gate scripts, regression routes, git host), ask for the rest. This skill uses DESIGN_FILE, TOKEN_DOCS, DS_ROUTE, GATES, MR_TOOL, DEPLOY_ORDER.

Core principles: agree on the plan before building, implement 1:1, every component gets a regression surface, finish with no loose ends. Three points where you MUST stop and wait for user confirmation: step 2 (plan), step 3 (off-system styles), step 7 (acceptance).

1. Skim + survey the current state + reuse audit

Skim the design with Figma MCP (get_design_context + screenshots): structure first, no parameter-digging yet. get_metadata tends to time out on large frames; prefer design context.
In parallel, map the existing implementation: related components, data flow, animation approach (an Explore agent works well).
Proactively enumerate reuse (don't wait to be asked): for every UI element in the design, look for an existing implementation and produce a reuse matrix (element → component → coupling → reuse strategy):
- What the design shows is usually the design frame of a real component. Assume it already exists, then try to disprove it, not the other way around.
- Coupling tiers: pure presentation (use directly) / light context (does it degrade gracefully without providers?) / heavy data coupling (data layer, runtime, store).
- A DS_ROUTE page or demo page rendering the component with mock props is direct evidence the presentation surface is already isolated.
- Heavy coupling ≠ no reuse: the default move is to extract the pure presentation core so the live caller, the regression page, and the new surface share it. A visual replica is the last resort, unless the user explicitly wants isolation.
Produce a diff list: structural differences between design and current state, the heavy-rework areas, and anything worth extracting as shared.

2. Plan review (STOP)

Present the plan for review: scope, component split/abstraction, animation handling, risks. Lead with a recommendation, not a menu. Wait for confirmation before step 3.

3. Extract detailed specs, block by block (conditional STOP)

Read each section with get_design_context + get_variable_defs, down to every gap, font size, and color token:

Map whatever fits the project's token/component system directly (consult TOKEN_DOCS).
Report every off-system style to the user, one by one (absorb as a new token? one-off hardcode? change the design?). If everything maps into the system, say so and continue without stopping.
Figma variable names ≠ code token names. Verify against the project's token files; never guess.

4. Full 1:1 implementation

Build to the confirmed plan. The target is pixel-level fidelity, not "roughly similar".
Respect the project gates; UI changes usually touch i18n keys, so handle them in the same pass.
Run a dev server and compare against screenshots as you go (multi-worktree port contention: read the actual port from your own server log; headless screenshots may render CSS fallbacks instead of WebGL/shaders, verify headed when needed).

5. Regression page (mandatory for component work)

When the change involves a reusable component (new, or visual/state changes to an existing one), create or update its DS_ROUTE/<component> page:

State enumeration follows the same standard as c2d2c:c2d: every variant axis (read from tv()/type unions, don't guess), every data shape (value/empty/fallback/truncation/loading/error), every interaction state (rest/hover/focus/selected/disabled + hover popovers).
The page renders from mock props, no live data. It is both the proof that the presentation surface is isolated and the state checklist for a future c2d2c:c2d run.
For heavily coupled components, extract the pure presentation core first (see step 1) so the live caller and the regression page share it: this kills hand-copied replicas that drift.
Pure page-level changes (no reusable component) may skip this step; say why.

6. Motion and state review

Walk every animation, transition, and hover/active/loading/empty state in the module:

Motion tokens compliant? (Projects usually have a motion ladder; inline springs/beziers are banned.)
Entrances/exits, scroll-linked behavior, reduced-motion fallbacks complete?
Fix what you find; report the motion inventory and verdict.

7. Screenshot acceptance (STOP)

What to capture: the regression page, full plus close-ups of key states; if there's no regression page, capture the real route. Use scripts/ds-shot.mjs bundled next to this SKILL.md (node ds-shot.mjs <route> [--port N] [--dark]) or manual browser capture.
Present them side by side with the Figma get_screenshot output; call out known deviations and why.
Wait for the user to pass acceptance before step 8. Requested changes loop back to step 4.

8. Cleanup + MR

Cleanup (no loose ends, no patches on top):

Delete every replaced implementation completely: component files, barrel exports, types, i18n source keys, unreferenced styles/tokens, portal mount points, dead constants. Leftover dead code misleads the next investigation.
Run GATES to confirm nothing dangles.

MR (runs automatically once acceptance passed and gates are green; do not stop to ask again):

Rebase onto the latest default branch.
Open the MR with MR_TOOL. Description template: Figma node link / acceptance screenshots / change summary / deletion list / gate results / DEPLOY_ORDER dependencies (e.g. "backend first").
Merge per project convention (e.g. GitLab: glab mr merge --auto-merge --yes with server-side rebase; queue auto-merge immediately on fast-moving mains).

Reporting rhythm

Steps 1-2 merge into one plan-review message (diff list + recommended plan).
Step 3 stops only for off-system styles; otherwise one line.
Step 7 is one acceptance message (side-by-side screenshots + deviation notes).
After step 8, close out: MR link, deletion list, gate results.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

scripts/ds-shot.mjs3 KB

Frequently asked questions

What does the D2c AI skill do?

Why use D2c on TypingMind?

Because you install it once and use it with any model. D2c is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install D2c in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/BIAsia/c2d2c/tree/main/skills/d2c. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use D2c?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with D2c?

As many as you like. As long as a model supports skills, you can use D2c with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the D2c AI skill free?

Yes. It is published on GitHub by BIAsia under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Concept Diagrams

NousResearch

OrganizationPopular

Generate flat, minimal educational SVG visuals as HTML.

Hyperframes

Render MP4/WebM videos from HTML compositions.

Ip As Logo

Design minimal cute IP mascot marks readable at 32px.

Mono Color

Generate one- or two-ink editorial print poster images.

Overview

Installation

Install in TypingMind

Install in another agent

Use it in TypingMind

Works with any AI model

Skill instructions

c2d2c:d2c — design to code, 1:1

Project parameters

1. Skim + survey the current state + reuse audit

2. Plan review (STOP)

3. Extract detailed specs, block by block (conditional STOP)

4. Full 1:1 implementation

5. Regression page (mandatory for component work)

6. Motion and state review

7. Screenshot acceptance (STOP)

8. Cleanup + MR

Reporting rhythm

Bundled files

Frequently asked questions

Related AI Skills

Set up your own AI workspace now