Sponsor Suno AI Music arrow_forward
Subagent

Claude Design Orchestrator

Parses claude.ai/design handoff bundles: validates schema, dedups proposed components against the codebase via component-search, reconciles tokens, and tracks bundle→PR provenance so design intent stays linked to shipped code.

Type
Subagent
GitHub stars
284
License
MIT
Repo last updated
Sep 27, 2026
Model
sonnet

What Claude Design Orchestrator is

Claude Design Orchestrator is a subagent published in the yonatangross/orchestkit repository on GitHub, which has about 284 stars. The repository describes itself as: “The Complete AI Development Toolkit for Claude Code. 106 skills, 36 agents, 171 hooks. Install `ork` for stable (v9.x), or `ork-alpha` for the v10 line, which ships daily.”

A subagent is a specialist assistant that Claude can hand part of a task to. It is a markdown file whose frontmatter sets a name, a description that tells Claude when to delegate, and optionally the tools and model it may use; the body becomes the subagent's own system prompt.

Because a subagent works in its own context, it keeps the main conversation focused: Claude can send a narrow job, such as a review or a specialised analysis, to Claude Design Orchestrator and get back a compact result.

How to install Claude Design Orchestrator

Claude Code

  1. Download claude-design-orchestrator.md from the repository.
  2. Save it to ~/.claude/agents/ to use it in every project, or to .claude/agents/ inside one project to share it through version control.
  3. Claude Code watches these folders, so the subagent is usually available right away. Ask Claude to use it by name, or @-mention it to make sure it runs.

Claude Cowork

  1. Cowork loads subagents through plugins. If the repository is packaged as a plugin marketplace, add it under Customize → Plugins → Add marketplace and install the plugin that contains this subagent.
  2. Otherwise, bundle the file into your own plugin's agents/ folder and upload it from Customize → Plugins.

New to extending Cowork? Our plugins guide and Customize guide explain how skills, plugins, and connectors fit together.

Inside the source file

An excerpt from plugins/ork/agents/claude-design-orchestrator.md, shared under the repository's MIT license. Read the full file on GitHub.

Directive

Own the seam between claude.ai/design and OrchestKit. When a handoff bundle arrives — URL or local file — parse it, validate it, deduplicate components against what already exists, and hand a normalized payload to downstream skills (design-import, design-ship). Treat the bundle as the source of truth for design intent; treat the codebase as the source of truth for implementation reality. Reconcile the two without surprising the user.

Consult project memory for prior bundles and their resulting PRs before starting. Persist bundle-to-PR provenance so future imports can detect when a component was previously generated from a different design version.

Read the handoff bundle in full before any scaffolding decision. Compare its declared design tokens against the project's actual tokens (Tailwind config, CSS custom properties, design-tokens.json). Run component-search across the codebase for every component the bundle proposes — never scaffold a duplicate of something that already exists.

When ingesting a bundle, run independent reads in parallel:

  • Fetch bundle (WebFetch if URL) + read existing tokens + glob existing components — all at once
  • For each component in the bundle, kick off a parallel component-search lookup
  • Memory queries (prior bundles, prior PRs from this design) — parallel to schema validation

The bundle declares what Claude Design generated. Don't re-derive what the bundle already says. Don't invent a token hierarchy when the bundle ships flat tokens. Don't refactor surrounding components just because the bundle touched them.

Agent Teams (CC 2.1.33+)

When running as a teammate:

  • Send normalized component manifest to frontend-ui-developer for implementation
  • Send extracted design tokens to design-system-architect for integration
  • Send component-match results to component-curator to update the registry
  • Use SendMessage to coordinate with design-context-extractor if the bundle is missing token metadata
  • Before any SendMessage to a peer outside your team, call ListAgents and address a listed name — never send to a guessed session name.
  • A reply to any message you send to another session is delivered to your PARENT session's conversation, not to you; send and move on, never wait for an answer. Cross-session messaging works on Bedrock, Vertex and Foundry and with telemetry disabled, so a provider or DISABLE_TELEMETRY=1 is not a reason to fall back to polling files.

Handoff Bundle Schema (observed format)

> Updated 2026-04-19 from real-world dogfood (#1399). The first iteration of Bet A assumed a JSON manifest with components[]. Reality is different — Claude Design ships an HTML-prototype tarball, not a structured manifest. The agent MUST work with the real format below; the prior JSON shape is obsolete.

Wire format

The URL https://api.anthropic.com/v1/design/h/ ?open_file= .html returns:

  • Content-Type: application/gzip
  • Body: a .tar.gz archive, typically 2–20 KB
  • Layout:
  <project-name>/
  ├── README.md              ← addressed to "CODING AGENTS: READ THIS FIRST"
  ├── chats/
  │   └── chat<N>.md         ← user ↔ assistant transcripts (load-bearing!)
  └── project/               ← only present once the assistant produced designs
      └── <Name>.html        ← self-contained HTML/CSS/JS prototypes

Key facts

  1. Prototypes are HTML, not a JSON components[] manifest. The README explicitly tells the coding agent to recreate them in whatever technology fits the target codebase, and not to copy the prototype's internal structure.
  2. Tokens are inline :root { --var: ... } CSS custom properties, not a separate JSON file. Extract from the HTML's block.

Before you install

  • Read the whole file first. Skills, commands, and subagents are instructions Claude will follow, so make sure they match what you want.
  • Check which tools, scripts, or MCP servers it uses. Local servers and scripts run with your permissions.
  • Try it in a test project or a copy of your files before pointing it at real work.
  • Pin the version you tested, and review changes before updating.
  • Watch for instructions that fetch web content or run shell commands; those are where prompt injection risks start. See our prompt injection guide.

FAQ

What is Claude Design Orchestrator?

Claude Design Orchestrator is a subagent for Claude Code and Claude Cowork from the yonatangross/orchestkit repository on GitHub. Parses claude.ai/design handoff bundles: validates schema, dedups proposed components against the codebase via component-search, reconciles tokens, and tracks bundle→PR provenance so design intent stays linked to shipped code.

How do I install Claude Design Orchestrator in Claude Code?

Download claude-design-orchestrator.md from the repository. Save it to ~/.claude/agents/ to use it in every project, or to .claude/agents/ inside one project to share it through version control. Claude Code watches these folders, so the subagent is usually available right away. Ask Claude to use it by name, or @-mention it to make sure it runs.

Can I use Claude Design Orchestrator in Claude Cowork?

Cowork loads subagents through plugins. If the repository is packaged as a plugin marketplace, add it under Customize → Plugins → Add marketplace and install the plugin that contains this subagent. Otherwise, bundle the file into your own plugin's agents/ folder and upload it from Customize → Plugins.

Is Claude Design Orchestrator safe to install?

It is a third-party community resource, not reviewed by Anthropic or this site. Read the source file first, check which tools and connectors it uses, and install only from sources you trust.

Similar resources

Browse all skills, subagents, and plugins →

Listing data comes from the public GitHub repository and was last checked in September 2026. Excerpts are © their authors and shared under MIT. This directory is independent and not affiliated with Anthropic or the resource's authors.