One library for every editor

One hub.
Every AI editor.

Write skills, rules, agents, and workflows once in ~/chain-hub — your single source of truth on disk. Every linked editor reads the same Markdown files after chain setup; Cursor, Claude Code, and Windsurf stay in sync without copy-paste. Open the dashboard (chain hub) or use the CLI.

Get started See the dashboard
You own the files
Link once, sync everywhere
Browse & install skills
Version with Git
Why Chain Hub

One mind.
Many editors.

Every AI tool stores your instructions, brand voice, and project context in a different place. You repeat yourself, updates drift apart, and your best guidance only works in the tool where you wrote it. Built for developers and power users who work across more than one AI coding tool.

Supported editors each expect skills and rules under their own paths — Chain Hub keeps one library under ~/chain-hub instead.

The old way

Fragmented

  • Manual copy-pasting

    Pasting brand voice and project rules into every new chat and every editor.

  • Inconsistent AI

    Claude knows your style; another tool still runs on generic defaults until you remember to update it.

  • Lost knowledge

    Expert prompts live in old chats or one-off config files — hard to reuse or version.

The Chain Hub way

One library

  • Update once, smarter everywhere

    Change a skill or rule in your hub — linked editors pick it up after setup.

  • Universal consistency

    Brand voice, skills, agents, and workflows stay aligned across the tools you use.

  • Permanent library

    Markdown you own on disk — version with Git when you want.

See the full editor list in How it works or read hub layout in the docs.

How it works

Three commands.
Then every editor.

Install the CLI, create ~/chain-hub, and run setup once. Your skills, rules, agents, and workflows live in one place — the dashboard and CLI keep them organized.

  1. Step 01

    Install the Hub

    Adds the chain command to your PATH (fixes common macOS npm permission errors).

    curl -fsSL https://www.chainhub.one/install.sh | bash

  2. Step 02

    Initialize your library

    Creates ~/chain-hub with protected core skills and empty buckets for your content. Open a new terminal tab after install if `chain` is not found yet.

    chain init

  3. Step 03

    Link your editors

    Detects installed IDEs and symlinks skills/, agents/, workflows/, and rules/.

    chain setup

Ready to install? Copy install, then init and setup — one command per step (see Install).

Symlink flow

~/chain-hub paths → editor folders

After chain setup, each detected editor reads from the same hub paths via symlinks — no copy-paste between config folders.

source
~/chain-hub/skills/debug/SKILL.md
chain setup links hub paths to each editor folder below
Cursor
~/.cursor/
Claude Code
~/.claude/
Nova
~/.claude/
JetBrains AI
~/Library/Application Support/JetBrains/*/aia/.agents/skills
Devin
~/.devin/
Antigravity CLI
~/.gemini/antigravity/
Gemini CLI (Legacy)
~/.gemini/
Mistral Vibe
~/.vibe/
Trae
~/.trae/
Kiro
~/.kiro/
Zed
~/.agents/skills
Universal
~/.agents/
Supported editors

Per-editor adapters

Click a card for symlink mappings and setup commands. Verify links anytime with chain status.

Status Live Beta Planned
Synced assets Skills Skills Reusable agent instructions from your hub skills folder, symlinked into each editor that supports skills. Agents Agents Specialist agent definitions from ~/chain-hub/agents/, linked into editors that load custom agents. Workflows Workflows Slash commands and multi-step workflows from ~/chain-hub/workflows/, mapped to each tool's command surface. Rules Rules Persistent project and global rules from ~/chain-hub/rules/, synced into editor-specific rules files. Tap a badge for asset types · Focus or hover a badge for what each letter means · click a card for symlink paths

Ready today

Adapters ship in the CLI today. Run chain setup once — installed tools get symlinks from your hub automatically.

9 editors

In beta

Early adapter support — feedback welcome while mappings stabilize.

4 editors

Missing an editor you rely on? Tell us which product and why it matters — it directly informs adapter priorities.

GitHub Discussions

No timeline guarantees — we read everything and use it to rank what we build next.

Editor Connection

~/.config/...

Symlink mappings
chain setup
$ chain status --check connections
✓ Infrastructure verified
Setup Command
chain setup --adapter ...
Synced Capabilities

Integration is fully automated via native symlinks — no manual config required.
Your hub, in the browser

See your whole library.
Fix drift in one place.

Run chain hub on 127.0.0.1: manage skill packs, filter by bucket and status, install from Discover, and open #system when symlinks or updates need attention — same library as the CLI.

Skills & packs

Toggle packs, filter by bucket (core, personal, community), and open any skill to edit markdown in the hub.

Discover

Browse the public catalog, install with chain add, and see what is new since your last sync.

Brain (Pro)

Memory rules plus the reflect → propose → apply pipeline; requires a Hub Pro license (install the Pro pack after activation).

Static preview — run chain hub for the live UI. Bundled skills: what ships with Chain Hub.

Get started
What ships with Chain Hub

14 ops skills on day one.
Add authoring when you need it.

14 protected operations skills mirror into ~/chain-hub/core/skills/ on every install. Add the 9-skill authoring pack with chain init --with-authoring or chain add — removable anytime.

Operations · 14 protected · core/skills/
Tier A — hub operations
chain-hub/ Hub map ~/chain-hub layout, registry buckets, IDE adapters
chain-setup/ NEW init, setup, status, first validate
chain-validate/ NEW Fix validate errors and registry drift
hub-sync/ NEW Git sync for ~/chain-hub across devices
hub-learn/ NEW Capture, reflect, improve proposals
chain-memory-rules/ NEW Durable memory rules for editors
registry-hygiene/ NEW Audit skills-registry.yaml buckets
skill-search-authoring/ NEW Descriptions and tags for discoverability
team-hub-standards/ NEW Shared team hub conventions
editor-adapter-troubleshoot/ NEW Broken symlinks and missing skills in editors
commit/ Conventional commit messages
debug/ Systematic root-cause debugging
code-review/ Structured multi-axis code review
find-skills/ Discover and install registry skills
Authoring pack · 9 skills · skills/ (optional)
chain init --with-authoring
Tier B — removable
hub-content-authoring/ Route to skill, rule, workflow, or agent authoring
create-skill/ Author SKILL.md with correct frontmatter
create-rule/ Create editor rules and project memory files
create-workflow/ Workflows that pass chain validate
create-agent/ Agents under ~/chain-hub/agents/
create-hook/ Cursor hooks and agent automation
migrate-to-skills/ Import legacy markdown rules as skills
update-cli-config/ Chain Hub CLI and persisted config
update-editor-settings/ Editor settings.json for VS Code family

Installed under ~/chain-hub/skills/ and registered in skills-registry.yaml (chain_hub bucket). Remove with chain remove <slug>.

Browse the public catalog on Registry or read installing content in the docs.

Add-on Early access

Same ~/chain-hub.
A deeper playbook.

Chain Hub Pro is an optional add-on on top of the open-source hub. Your library still lives in one place — ~/chain-hub — with no second copy on disk. Pro unlocks Brain (Memory + Pipeline), adds a private skill pack (44 skills, 15 agents, 19 workflows, pack 0.16.1), and extends Design System with Pro presets and generators.

  • Get your key — checkout, beta approval, or Account
  • Activate — chain pro activate or Settings → Pro unlocks Brain
  • Install pack — chain pro install-pack or the hub Dashboard
Learn about Pro Pack inventory (docs)

Design Systems

Pick a visual direction and export tokens to your repo

Hub Brain

Capture learnings and turn them into skills you keep

Ship gates

Review, test, and ship with built-in checklists

Modern stack

Astro, Next.js, and React workflows with version-aware docs

Go deeper

Details live in the docs

The homepage stays focused on the core story. Sync, CLI tables, hub anatomy, and learning pipelines are documented step by step.

Quick start

Install in three
terminal steps.

Run the install script (handles macOS npm permission issues), initialize ~/chain-hub, and link editors. Use one terminal block per step — pasting every line at once often fails because chain is not on PATH until you open a new tab or reload your shell.

Copy buttons on steps 2–3 add ~/.npm-global/bin for the same session. Or open a new terminal tab after install and run the commands as shown. Optional: chain init --with-authoring.

Step 1 — Install
curl -fsSL https://www.chainhub.one/install.sh | bash

# or: npm install -g chain-hub · if chain is missing: new tab or source ~/.zshrc

Step 2 — Init hub
chain init

# optional: chain init --with-authoring

Step 3 — Link editors
chain setup
Scanning for installed IDEs…
✓ Devin (skills) → ~/.codeium/windsurf/skills
✓ Devin (workflows) → ~/.codeium/windsurf/global_workflows
✓ Devin (rules) → ~/.codeium/windsurf/memories/global_rules.md
✓ Cursor (agents) → ~/.cursor/agents
Hub linked. Core + connections ready.

Example output only — not copied.

Optional — Custom hub path
chain config set chain_home ~/my-hub
FAQ

Common
questions.

Quick answers about setup, privacy, and how Chain Hub fits your editors. Everything starts collapsed — open one question at a time.

Getting started

Install failed with a permission error on macOS?

That usually means npm cannot write to /usr/local (common with the Node installer from nodejs.org). Do not use sudo npm install.

Run our install script instead — it configures a user-owned npm directory when needed, then installs Chain Hub:

curl -fsSL https://www.chainhub.one/install.sh | bash

Then chain init and chain setup. More detail in Troubleshooting and Quickstart.

`chain` not found right after install?

The installer may add ~/.npm-global/bin to your shell profile. Terminals that were already open keep the old PATH until you reload.

Open a new terminal tab or window, or run source ~/.zshrc (bash: source ~/.bashrc), then run chain init and chain setup.

See Troubleshooting if which chain still fails.

Do I need to be a programmer to use this?
Not at all. Chain Hub uses a CLI for setup, but your knowledge lives in plain Markdown — the same format you might use in Notion or Obsidian. After chain init and chain setup, you mostly edit text files in your hub folder.
How do I update my knowledge?
Open the files in ~/chain-hub, edit, and save. Every linked editor picks up the change on the next session — no copy-pasting between tools.
What ships with chain init?

Every install gets protected operations skills under core/skills/; you can add the optional authoring pack with chain init --with-authoring.

See What ships with Chain Hub for the full Tier A and Tier B lists.

Privacy & pricing

Is my data safe and private?
Yes. Chain Hub is local-first: skills, rules, agents, and workflows stay on your machine under ~/chain-hub. Nothing is uploaded to a Chain Hub cloud. You own the files and choose if and how you back them up (for example Git sync in the docs).
Does it cost anything?
Chain Hub is open-source and free on npm. Install it for personal or professional use without a subscription.

Editors & concepts

Which AI tools does this work with?

Run chain init then chain setup to symlink ~/chain-hub into each editor Chain Hub detects on your machine.

See Supported editors for live, beta, and planned adapters with per-tool paths.

What is a Skill vs an Agent vs a Workflow?

Skills, rules, workflows, and agents are different markdown types in your hub — each with its own path and who runs it (assistant, you, or a subagent).

See Hub layout in the docs for definitions and on-disk paths.

Need more detail? Read the docs →