Foundry — Design skill for Claude Code
Turn a goal into finished, tested and reviewed work.
How to install Foundry
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open louiskhenghao/Foundry and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Foundry does
Turn a goal into finished, tested and reviewed work. Foundry runs locally and orchestrates Claude Code or OpenAI Codex through one web UI, from planning to delivery. Uses your Claude or ChatGPT subscription; no inference API key required.
Alternatives in Design
- Pipelines — Pipelines: Versatile, UI-Agnostic OpenAI-Compatible Plugin Framework 2.4k ★
- Designer Skills — Designer Skills Collection: agentic skills, commands, and plugins for design — from research to systems, UI, i 2.3k ★
- Vox Director — Turn one topic into a finished Vox-style paper-collage explainer/ad video — automated end to end on Atlas Clou 2.1k ★
README
Foundry
**Write the goal in one sentence. Get back finished, checked, reviewed work.**
Foundry runs on your own computer and drives your Claude Code or Codex CLI from a web page:
it asks what only you can decide, shows you the plan once, then builds, tests, reviews and delivers on its own.
English · [中文](./README.zh.md)
[Quick start](#quick-start) · [How it works](#how-it-works) · [User guide](docs/guide/) · [Docs](docs/)
Why Foundry
Claude Code and Codex can carry out a request. Keeping a real piece of work on track is the part left to you: saying what you actually mean, splitting it up, checking every step, noticing when it drifts, and not letting it burn your plan. Foundry does that part.
- One sentence in
"Add CSV export to the orders page." "A landing page for our studio." "Posters for the new drink." - It asks before it plans
It reads your repository first and asks only what the code can't answer — each question with a recommended answer. - One plan to approve
The Brief: what it understood, the tasks, how the result will be checked, what it may cost. Edit anything, then approve. - You see it before it's done
At milestones it pauses, starts a live preview and waits for your look. - Checked, not guessed
Every task has acceptance checks; the whole result is reviewed against them before it is handed over. - Your code stays yours
Foundry runs locally and uses the selected CLI's account: Claude login or ChatGPT sign-in for Codex. Agent inference needs no API key. Work happens on its own branch; your checkout stays unchanged while it runs. - You control the spend
Independent model presets pick which model does ea
Related Skills
Claude Planning Skills
Four interlocking Claude skills: brainstorm a design, plan it, build it in reviewed batches, with a skeptic th
Carousel Lite
Free Claude Code skill: turn any topic into a finished Instagram carousel. AirDrop-ready PNGs, one clean style
Generating Slides
Claude Code skill: turn one sentence or a document into a projector-grade HTML slide deck — single file, zero
Ppt
PPT slide generation workflow: init -> requirement research -> material collection -> outline planning -> plan
UX Pgm Skills
Agentic skills, commands, and plugins for UX program management — from planning and stakeholders to delivery,
Insight To Quality
Structured discovery-to-delivery workflow built around discovery, system design, system mapping, implementatio
Related Agents
External Reviewer
Gets an independent second opinion on a finished resume or application from a different model (OpenAI Codex CL
Video Studio Team Lead
Lead of the AI video production team. Orchestrates storyboard writing, visual asset generation, Remotion engin
CLI Subagents Codex
Runs a task in OpenAI Codex (codex exec) on one of the user's ChatGPT accounts (CLI Subagents). The worker rea