Skip to content

In the making

Your whole agent setup, in one place.

Skills, tools, plugins, automations and agent definitions.

We’ll only email you about early access.

The Setup Craft mascot lying back on a sunlit cloud, a bar across his shoulders.

People investing in their setups

Real posts on the skills, tools and workflows behind their agents.

  • Design

    • Plugins
    • Tools
    • Skills
    On medium.com

    Nathan Curtis

    Founder, EightShapes; maker of the Specs Figma plugin

    “People encounter my specs tooling often first through the Specs Figma plugin that’s been around for years. … Commands fetch files, scan for components marked READY_FOR_DEV and generate a library’s worth of specs in less than a minute.”

  • Creators

    • Skills
    On jennakutcherblog.com

    Jenna Kutcher

    Creator, host of the Goal Digger podcast, founder of a creator education business

    “Together, they help me save 10 hours a week in my business. … Before the Repurpose Skill, I used to spend hours turning one piece of content into several social posts, an email and Pinterest Pins. Now I spend twenty minutes.”

  • Development

    • Plugins
    • Tools
    On openai.com

    Ryan Lopopolo (OpenAI)

    Member of Technical Staff, OpenAI

    “Over the past five months, our team has been running an experiment: building and shipping an internal beta of a software product with 0 lines of manually-written code. … We estimate that we built this in about 1/10th the time it would have taken to write the code by hand. … Logs, metrics, and traces are exposed to Codex via a local observability stack that's ephemeral for any given worktree.”

  • Legal

    • Skills
    On claude.com

    Darryl Sladden

    Staff Product Manager for AI, Box

    “Using the Skills API, Box went from concept to customer-facing capability in weeks. … A member of Box's internal team tested the agent on a real contract, asking it to show redlines for changing the governing jurisdiction from New York to California. … “We just waited two minutes, and it output the entire document,” Darryl said.”

  • Research & Science

    • Plugins
    • Agent Definitions
    On anthropic.com

    Matthew Schwartz

    Professor of Physics, Harvard; visiting researcher at Anthropic

    “This iterative process generated a set of software and scientific protocols, which I call BootLoops. … Running many projects at once (36 manuscripts in 18 fields with 19 coauthors over three months, out of some 400 candidate problems) takes a lot of coordination, both for me and for Claude. … I have separate sessions for each project, plus a master session that coordinates the others, allocates compute, and validates results.”

  • Design

    • Agent Definitions
    • Skills
    On github.blog

    Eric Bailey

    Accessibility / design systems, GitHub

    “For purpose number two, the accessibility agent is set to automatically evaluate changes that modify our front-end code. To date, the agent has reviewed 3,535 pull requests, with a 68% resolution rate.”

  • Business

    • Tools
    On datadoghq.com

    Datadog (Bowen Chen, Yosra Chelbi, Nader Khalil, Dmitrii Ulianin)

    Datadog engineering blog

    “To do this, our AI developer experience team built an agentic evaluation platform that provided teams with a self-service workflow to write, test, and add agent evaluations. … After running different models across over 140 different evaluations, we calculate their average cost and performance score on a nightly basis. … When monitoring the cost impact of this migration over the past month, over $687,000 in cost savings was attributed to this change.”

  • Marketing

    • Skills
    • Tools
    • Automations
    On X

    Tim Soulo

    CMO of Ahrefs

    “Ryan Law, our Director of Content at @Ahrefs, built a fully automated Claude Code workflow that creates publish-ready blog articles in just 8 mins. It runs on 23 custom skill files + Ahrefs MCP.”

  • Consulting

    • Plugins
    On claude.com

    Hyosup Bae (LG CNS)

    Head of Build Center, LG CNS

    “To deliver that, the Build Center, LG CNS's 200-engineer technology delivery organization, built an engineering orchestration layer around Claude Code. The harness connects legacy analysis, business logic extraction, code generation, testing, verification, and quality management. … The project achieved a 99.1% API mapping completion rate, with 2,888 of 2,913 APIs converted.”

  • Development

    • Tools
    • Agent Definitions
    On claude.dev

    Anthropic

    Maker of Claude

    “Once Claude can measure something, it can make it faster. … This August, we made the core user experience of claude.ai and the Claude desktop app about 3x faster in a two-week sprint. … We asked Claude to analyze usage data through the Datadog MCP server.”

  • Business

    • Plugins
    • Agent Definitions
    On hacks.mozilla.org

    Mozilla (Brian Grinstead, Christian Holler, Frederik Braun)

    Firefox engineering and security leads, Mozilla

    “We began with small-scale experiments prompting the harness to look for sandbox escapes with Claude Opus 4.6. … We fixed a total of 423 security bugs in releases in April.”

  • Sales

    • Skills
    • Tools
    On posthog.com

    Leonhard Prinz

    Technical Account Manager, PostHog

    “My entire opportunity review is now a skill in PostHog Desktop (or Claude) … This has condensed that operational time by 80%.”

  • Marketing

    • Plugins
    • Skills
    • Agent Definitions
    On seerinteractive.com

    Wil Reynolds

    Founder & CEO, Seer Interactive

    “We've got one skill for client's traffic drops, one for competitive research, dozens of these across the company … I gave that feedback in a 9-minute recording and an agent picked up my feedback and fixed it the same day … Git commits sync to our Claude plugins company-wide.”

  • Design

    • Skills
    • Tools
    • Agent Definitions
    On uber.com

    Ian Guisard (Uber)

    Leads design of Uber's Base design system

    “uSpec is an agentic system that connects an AI agent in Cursor® to Figma® through the open-source Figma Console MCP. … A full screen reader spec covering 3 platforms generates in under 2 minutes. A system with hundreds of components that previously required months of spec-writing can generate complete specs in days.”

  • Finance

    • Plugins
    • Agent Definitions
    On openai.com

    Lucas Baker

    Head of LLM R&D, Jump Trading

    “Where we used to require frequent human guidance and intervention, we can now focus fully on defining a secure and well-monitored environment with clear goals and letting the agents find their own way. You can define something that needs to run for days—it needs to pull from many data sources, it needs to make those subtle calls about what is important and what is not, and it needs to interrelate everything—to create a comprehensive analysis that actually works now.”

  • Consulting

    • Skills
    • Agent Definitions
    On josephthacker.com

    Joseph Thacker (rez0)

    AI security researcher and consultant; bug bounty hunter

    “This is the story of JD (xssdoctor) and I building a hackbot which found 126 bugs in the last 5 months. … Recon skills would map the attack surface. … 88 of 126 findings (70%) were rated High or Critical.”

  • Creators

    • Tools
    • Agent Definitions
    On robinsloan.com

    Robin Sloan

    Novelist (Mr. Penumbra's 24-Hour Bookstore, Moonbound)

    “I got home, and I told it, and I cooked dinner, and by the time I went to bed, I had my new notes app, native macOS. … As the capstone for my AI summer, I built myself an email client!”

  • HR

    • Plugins
    • Tools
    On LinkedIn

    Preetham Phirangi

    HR Business Partner

    “Today I reviewed 370 applications, shortlisted 6 candidates, in under 20 minutes using an AI-powered tool I built by myself with a free Google AI subscription using Antigravity.”

  • Research & Science

    • Skills
    On X

    Jim Fan

    Director of Robotics and Distinguished Scientist, NVIDIA (GEAR lab)

    “Today, we give robots a /skills library that self-evolves and compounds indefinitely! … ASPIRE achieves up to ~10x cut in "transfer learning” tokens (yes, tokens are the new unit of *training* compute ;) … Check out our gallery of 150+ tasks and 90+ skills the robots taught themselves, all on the website!”

  • Marketing

    • Tools
    • Automations
    On dejan.ai

    Dan Petrovic

    Founder, DEJAN AI

    “We spent 2.71 Billion tokens teaching a machine to reverse-engineer how AI search ranks pages. … The engine drove a page to the number one position in 1,554 of 2,246 runs, about 69% of the time.”

  • Creators

    • Plugins
    • Skills
    On macstories.net

    Federico Viticci

    Founder and editor-in-chief, MacStories

    “In my tests, Shortcuts Playground was able to one-shot dozens of simple shortcuts based on Apple’s built-in actions as well as complex shortcuts that involved web APIs, advanced conditional logic, SSH and shell scripting, and more. … Once I reached 1,000 shortcuts created and tested with different agents and realized the skill was good enough for a 1.0, I decided to wrap up my work on it for the time being.”

  • Business

    • Skills
    • Agent Definitions
    On engineering.fb.com

    Meta (Shaurya Sengar, Jason Nawrocki, Jay Shah, Prashant Kommireddi)

    Meta Engineering blog

    “We solve this with composable procedures we call recipes. … After restructuring into recipe-driven stages, each query touches only a small, targeted subset, cutting tokens consumed per turn by around 80%. … Days to minutes reduction in individual assessment time.”

  • Customer Support

    • Tools
    • Agent Definitions
    On monzo.com

    Jamie McDonald-Gibson, Joost van Oorschot, Robin Dhamankar and Tom Leitch

    Staff Engineer, Staff Backend Engineer, Distinguished Engineer and VP Engineering, Monzo

    “We introduced intent-gated customer context, such as transaction history, so the model could use relevant information without being overloaded by unnecessary data. This led to a 10 percentage point improvement in resolution for transaction queries. … Instead of relying on brittle mocks, we built a simulated environment where tool calls operate against realistic state.”

  • Sales

    • Agent Definitions
    • Tools
    On saastr.com

    Jason Lemkin

    Founder of SaaStr, the largest B2B SaaS community

    “Amelia built a renewal agent on top of 10K in about half a day. … With the agent, 20 to 30 custom decks went out.”

Post placeholder

Real posts will open here with the full text, who wrote it and a link to the original.

Agents are already extremely capable.Your setup either amplifies it or reduces it.

Clear instructions, the right tools and a little structure turn what an agent can do into work you can count on.

Understand the basics of an agent setup

The five pieces

Five pieces. One setup.

The parts most AI setups are made of, and what each one does.

  • Skills

    Teach your agent how you like the work done.

    A skill teaches your agent how to handle a specific task. Think of it like a recipe: you set the steps, technique, and taste so the work turns out right every time.

    • On demand: An agent can hold multiple skills and use only what is relevant.
    • Rule of thumb: Start with common skills, but hand-write the ones that matter most.

    For example

    A research skill that weighs both sides of a hard decision with clear pros and cons.

  • Tools

    Give your agent more ways to help.

    Tools let your agent look beyond the chat. It can check things for itself, find what matters, and take action, instead of waiting for you to bring it all the information.

    • Look: Find out what’s happening, whether it’s in your messages, calendar or files.
    • Do: Use what it learns to send a reply, make a change or take the next step.

    For example

    An agent looks through customer messages, spots a question that keeps coming up, and drafts a helpful reply.

  • Plugins

    Add something new, or change how it works.

    A plugin is an add-on that gives your agent a new ability or changes how it already works. Install it once and it stays ready for every task after that.

    • Adds: Brings in something new, such as a quick command or a helper for one job.
    • Changes: Adjusts the way your agent works so it fits how your team already does things.

    For example

    A review plugin that adds a quick review step and a checklist for every change your agent makes.

  • Automations

    Work that runs on its own.

    An automation is a whole job your agent does on its own, from start to finish. It can run several steps in order, such as gathering information, checking it and sending the result, without you clicking through each one.

    • On a schedule: Starts at a set time, like every weekday at nine, so the routine never gets forgotten.
    • On a trigger: Starts the moment something happens, like a new customer email or a finished report.

    For example

    When a customer emails asking for a refund, the agent checks their order, drafts a reply with the right amount and leaves it ready for you to approve.

  • Agent Definitions

    Turn a general agent into a specialist.

    An agent definition is a job description for one agent. A general agent knows a little about everything, which makes it unpredictable on a specific job. A definition narrows it to one role and tells the agent how to approach that work, so it gives the same kind of result every time.

    • Role: Sets what the agent is for and how it should approach its work.
    • Scope: Limits what it can see and change, so a mistake stays small and easy to catch.

    For example

    A design-review agent that checks each new screen against your brand colours and spacing, flags anything that looks generic, and fixes it in the design files. It can read and edit design files, but it never touches the code.

What we’re building

One home for your agent setup

Setup Craft is early. Here’s what we’re working on.

In the making

Product

A library of setups that work.

Skills, tools, automations, agent definitions and plugins, plus the thinking behind great setups, gathered from every way of working and every field. Find what fits you and install it in the harness you already use.

Every piece of a setup

  • Skills
  • Tools
  • Plugins
  • Automations
  • Agent Definitions

Installs into

  • Claude Code
  • Codex
  • OpenCode
  • Pi
  • DeepSeek harness
  • and more

Later: A marketplace: explorers subscribe monthly, and creators earn from the views and installs their work gets.

Services

We build the setup with you.

The right setup changes how much your agents can do, but few busy teams have the time to learn it all. So we offer two ways in.

Talk to us
  • First Setup

    Quick win

    A setup built around how your team really works.

    We study how each person works and how the team works together, then build a setup that fits both. You see results fast, and it carries you for a while.

  • Last Setup

    We recommend

    The setup, and the know-how to run it yourselves.

    We build your setup and teach your team every part of it: what each piece does, how to use it, and how to refine it as your work changes. It takes longer, and it leaves you AI native on your own, so you never need to hire us again.

Help shape Setup Craft

What would make this useful for your setup?

No account needed.

Only if you’d like us to reply.

Plans may change as we learn. These are aims, not shipped features.

Ready to build a stronger setup?

Join the waitlist for early access.