SkillFoxx

ai-registry (aireg) CLI

Open registry format for skills and agents plus the aireg CLI: pinned versions, rollback and sync for Claude Code, Codex, Cursor.

  • 179 ★ on the repository
  • Last commit
  • Apache-2.0
CLI

Extended access

Why this level

  • The installer runs a script from the network and drops files into your agent directories.
  • Through a registry a team receives and updates skills that agents then run.
All reasons and checks

atlanai/ai-registryAuthor:atlanai

Install

Manual install

curl -fsSL https://aireg.sh/install | sh

For macOS and Linux; on Windows use `irm https://aireg.sh/install.ps1 | iex`. Read the script first.

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • The installer runs a script from the network and drops files into your agent directories.
  • Through a registry a team receives and updates skills that agents then run.

This is third-party code. Review the repository files before installing.

What it does

ai-registry is an open standard for a registry of AI artifacts, meaning skills and agents, with one source of truth for every team and every coding agent. A registry stores versions, owners and approvals in a defined file layout, so it can be built in an ordinary git repository or object storage and read by any client. The aireg client works with registries in folders and git repositories: it creates them, publishes to them and installs from them into Claude Code, Codex and Cursor. `aireg add` lays skills out in agent directories and records exact versions in `.registry/manifest.lock.yml`. It also installs from a plugin marketplace, a site Agent Skills index or a folder. Unlike a plain skills repository, versions are pinned and verified, every machine gets identical bytes, a version can be rolled back or yanked for everyone and all agents sync from one record. An `ai-registry` skill lets the agent set up a registry on request.

Who it is for. Teams and organizations where dozens of people use different coding agents and need one vetted set of skills and agents with version control.

Good fit when

  • Colleagues install different copies of one skill and nobody knows who has which.
  • A faulty skill must be pulled from everyone at once and rolled back.
  • An organization wants its own registry on an existing git repository with no separate service.
  • One skill set must reach Claude Code, Codex and Cursor at once.

Not a fit when

  • You just need a couple of skills for yourself: `npx skills add` is enough.
  • You want a ready cloud catalog with login and search: the project gives a format and a CLI.
  • The registry must live on a service with an API and not in files: the CLI reads folders and git repositories.

Example request

Set me up with ai-registry: https://github.com/atlanai/ai-registry. Ask before installing anything, then try the example registry.

Limitations

This is a specification and a client and no ready service. Today aireg reads registries in folders and git repositories; other storage is described by the format but wider client support is not promised. The `curl -fsSL https://aireg.sh/install | sh` installer runs a script, so read it first or pick a specific release. On Windows use `irm https://aireg.sh/install.ps1 | iex`. The installer asks about the aireg skills when run in a terminal. You must build and maintain a registry yourself, including approvals and write access.

How to disable. Delete the `aireg` binary, the skill directories it laid out and the `.registry/manifest.lock.yml` file from the project. A registry in a git repository stays and is deleted separately.

FAQ

How do I install aireg?

On macOS and Linux with `curl -fsSL https://aireg.sh/install | sh`, on Windows with `irm https://aireg.sh/install.ps1 | iex`. Check with `aireg --version`.

How do I try the example registry?

Create a project with `git init`, then `aireg registry add github.com/atlanai/ai-registry --path examples/acme-registry`, `aireg add acme/engineering` and `aireg list`.

Where are versions recorded?

In the project `.registry/manifest.lock.yml`: it pins the exact version of each installed skill and agent.

How is a registry better than a plain skills repository?

It pins and verifies every version, gives every machine identical bytes, rolls back or yanks a version for everyone and syncs agents from one record.

Can an agent set up a registry itself?

Yes, the repository ships an `ai-registry` skill. The agent asks before installing aireg and its skills, then shows an example and helps create a registry.

Discussion

No questions or reviews yet.

CLI tools

A registry and arctl CLI for MCP servers, agents, skills and prompts: publish, run and configure clients in one place.

CLIExtended access539

A CLI that installs shared skills, commands, subagents, hooks and MCP servers into eight AI coding tools in their native layouts

CLIExtended access91

A skill and Python library: four agent decision-quality indicators, a repeated-write guard and a circuit breaker for agent chains

SkillExtended access309
Official

A self-hosted MCP and A2A gateway with identity provider integration, access control and observability for enterprise agents

MCP serverBroad access3.9K

gitai is an open source-control server for hundreds of agents: landing by compare-and-swap, server-side history and the graft CLI.

CLIExtended access1K

A self-hosted enterprise skill registry: publishing, search, versions, tags and tokens, MinIO storage, plus the @astron-team/skillhub CLI

CLIBroad access5.4K

README in short

The README explains how a registry differs from a skills repository and a marketplace and shows the path from installing aireg to a first registry. The project holds a format specification with JSON schemas and the aireg client. A project roadmap is described.

README badge

Report a problem