Managing AI Artifacts for Your Team

Create a Git-hosted artifact repository to distribute skills, AGENTS.md, and Knowledge Base content across your team with pair-cli.

A walkthrough for teams who want to organize and distribute AI artifacts — skills, AGENTS.md, and KB content — via a Git repository, independent of your development workflow. This is not about adopting pair's full product lifecycle management; it's about using pair-cli to version and share artifacts across your team and multiple AI assistants. By the end, every teammate can install and update your organization's artifacts with a single command.

Prerequisites

pair-cli installed (pair-cli --version) — Install pair-cli if you haven't already
Git installed (git --version)
A Git hosting account (GitHub, GitLab, Bitbucket, or self-hosted)

What you'll build

An artifact repository with skills, AGENTS.md, and a manifest
A versioned distribution workflow using Git tags
A standard config.json that your team uses for consistent installs
A one-command install for teammates: pair-cli install --source <git-url>#v1.0.0 --config path/to/config.json
Private repo auth for CI/CD environments

Estimated time

~30 minutes.

Step-by-step instructions

1. Choose a prefix

Pick a short, unique prefix for your organization (e.g., acme). This prevents name clashes when skills from multiple sources are installed in the same project.

acme-capability-verify-quality
acme-process-implement
^^^^
prefix

2. Create the artifact repository

mkdir acme-ai-artifacts && cd acme-ai-artifacts
git init

3. Choose your layout

Flat layout — all skill directories at the root:

acme-ai-artifacts/
  AGENTS.md
  manifest.json
  acme-capability-deploy/
    SKILL.md
  acme-process-onboard/
    SKILL.md

Folder-organized layout — grouped by type, flattened on install:

acme-ai-artifacts/
  AGENTS.md
  manifest.json
  .skills/
    capability/
      deploy/
        SKILL.md
    process/
      onboard/
        SKILL.md

The folder layout is easier to maintain at scale. pair-cli's flatten + prefix transforms handle the renaming.

4. Create a team `config.json`

The config.json defines how pair-cli distributes artifacts. Create it in your team's shared location (Git repo, wiki, or docs folder) or have each developer create one locally.

For the folder-organized layout, start with a single assistant:

{
  "asset_registries": {
    "skills": {
      "source": ".skills",
      "behavior": "mirror",
      "flatten": true,
      "prefix": "acme",
      "description": "Acme team skills",
      "targets": [
        { "path": ".claude/skills/", "mode": "canonical" }
      ]
    }
  }
}

This delivers all skills to Claude. When your team adopts other assistants (GitHub Copilot, Cursor, Windsurf), update targets:

{
  "asset_registries": {
    "skills": {
      "source": ".skills",
      "behavior": "mirror",
      "flatten": true,
      "prefix": "acme",
      "description": "Acme team skills",
      "targets": [
        { "path": ".claude/skills/", "mode": "canonical" },
        { "path": ".github/skills/", "mode": "symlink" },
        { "path": ".cursor/skills/", "mode": "symlink" },
        { "path": ".windsurf/skills/", "mode": "symlink" }
      ]
    }
  }
}

For the flat layout, omit flatten and prefix since directory names already include the prefix.

Where to store config.json:

Option 1: In your project root (version-controlled with your code)
Option 2: In a shared team location and pass via --config flag at install time

Advanced: Use the transform property to customize files per AI tool (e.g., Claude-specific vs Cursor-specific AGENTS.md). See Integrations for details.

5. Write AGENTS.md

Create a minimal AGENTS.md that documents how to install your artifacts. If you're storing config.json in the repo, reference it:

# AGENTS.md
 
This project uses Acme AI artifacts. Skills are installed to `.claude/skills/`, `.github/skills/`, and `.cursor/skills/`.
 
## Available Skills
 
- `/acme-capability-deploy` — Deploy to staging/production
- `/acme-process-onboard` — Onboard a new team member
 
## Setup
 
Install with:
 
\`\`\`bash
pair-cli install --source https://github.com/acme/ai-artifacts.git#v1.0.0 --config acme-config.json
\`\`\`
 
Or if using your project's config:
 
\`\`\`bash
pair-cli install --source https://github.com/acme/ai-artifacts.git#v1.0.0 --config ./pair-config.json
\`\`\`

Customize per AI tool: You can generate different AGENTS.md versions for Claude, GitHub Copilot, Cursor, etc. using the transform property in your config. See AGENTS.md Customization to create tool-specific guidance and onboarding instructions for each assistant.

6. Create `manifest.json`

{
  "name": "acme-ai-artifacts",
  "version": "1.0.0",
  "description": "Acme team AI artifacts",
  "author": "Acme Engineering"
}

7. Commit, tag, push

git add .
git commit -m "feat: initial artifact package v1.0.0"
git tag v1.0.0
git remote add origin git@github.com:acme/ai-artifacts.git
git push -u origin main --tags

8. Install on a teammate's machine

Pass the config file with --config:

# HTTPS (public repo)
pair-cli install --source https://github.com/acme/ai-artifacts.git#v1.0.0 --config ./pair-config.json
 
# SSH (private repo — uses SSH keys)
pair-cli install --source git@github.com:acme/ai-artifacts.git#v1.0.0 --config ./pair-config.json

The CLI shallow-clones the repo, reads the config file you provide, and distributes the content to the configured targets.

9. Private repos

SSH (git@…) — works if the user has SSH keys configured. No extra configuration needed.

HTTPS + token — for CI/CD or environments without SSH keys:

export PAIR_GIT_TOKEN=ghp_your_token_here
pair-cli install --source https://github.com/acme/ai-artifacts.git#v1.0.0 --config ./pair-config.json

In a GitHub Actions workflow:

- name: Install AI artifacts
  env:
    PAIR_GIT_TOKEN: ${{ secrets.ARTIFACTS_TOKEN }}
  run: pair-cli install --source https://github.com/acme/ai-artifacts.git#v1.0.0 --config ./pair-config.json

10. Update workflow

When you update your artifacts:

Make changes to the artifact repository
Bump the version in manifest.json
Commit, tag, and push:

git commit -am "feat: add deployment skill v1.1.0"
git tag v1.1.0
git push origin main --tags

Teammates update with:

pair-cli update --source https://github.com/acme/ai-artifacts.git#v1.1.0 --config ./pair-config.json

11. Scale to multiple assistants

As your team grows and adopts new assistants, simply update the targets array in config.json. No changes needed to the skills themselves.

Before (only Claude):

"targets": [
  { "path": ".claude/skills/", "mode": "canonical" }
]

After (Claude + GitHub Copilot + Cursor + Windsurf):

"targets": [
  { "path": ".claude/skills/", "mode": "canonical" },
  { "path": ".github/skills/", "mode": "symlink" },
  { "path": ".cursor/skills/", "mode": "symlink" },
  { "path": ".windsurf/skills/", "mode": "symlink" }
]

Once updated, teammates run:

pair-cli update --source https://github.com/acme/ai-artifacts.git#v1.1.0 --config ./pair-config.json

The new locations are populated automatically. Existing assistants' skill directories remain untouched.

You can also host the package as a ZIP on any HTTP server, network share, or artifact registry. See Install from URL or Path.

What you've learned

How to structure an artifact repository with skills, AGENTS.md, and manifest
How to create a team-standard config.json for consistent distributions
How to use flatten + prefix for organized skill directories
How to version and distribute artifacts via Git tags
How to handle private repo auth with SSH keys and PAIR_GIT_TOKEN
How to deliver to multiple AI assistants with a single install command

Next steps

Team Setup — Full team onboarding with pair
Install from URL or Path — Alternative distribution methods
Configuration Reference — Full config.json reference
Skills Concept — How skills work

Managing AI Artifacts for Your Team

On this page