# HappySkills

> The package manager for AI agent skills: dependencies, a lock file, sync and team sharing, all by talking to your agent. I built it, and I use it every day.

- Verdict: My default
- Vendor: HappySkills (founded by Nicolas Dao)
- Layer: [Build and test](https://cloudlesslabs.com/stack/layers/dev/index.md). Job: Package manager for agent skills
- Last reviewed: 21 October 2026, by Nicolas Dao (Cloudless Labs, Sydney)
- Web page: https://cloudlesslabs.com/stack/happyskills/. The whole stack in one file: https://cloudlesslabs.com/stack/brief.md

My own product, so the bias is on the table: I founded HappySkills, and it's the tool I use most after Claude Code itself.

Skills are how you turn an agent from a clever generalist into someone who knows how your company works: your release process, your review checklist, your way of writing a spec. They're plain folders with a SKILL.md, and that's the problem. Once you have dozens, they get copied between projects, edited in one place and not the other, and nobody knows which copy is right. The Agent Skills standard defines the file; it says nothing about versions, dependencies or sharing.

HappySkills is the missing layer, the npm of skills, without changing the format:

- **Dependencies.** A skill declares the skills it builds on, with version ranges. One install resolves the whole tree, flat, one version of each, and stops with the chain that caused it when two ranges can't agree. I introduced this because the spec has no dependency field, and big skills only stay reliable when they're split into small ones.
- **A lock file.** `skills-lock.json` records the exact version, commit and SHA-256 hash of every skill. Commit it, and `npx happyskills install` rebuilds the same setup anywhere, refusing to install anything whose hash doesn't match.
- **Sync without losing work.** It knows whether each skill is clean, modified, outdated or diverged. When you've edited a skill and the registry has moved on, it does a three-way merge from the version you started from, with git-style conflict markers, instead of overwriting your work.
- **Kits.** A whole set of skills installed under one name, with the versions pinned in your lock file.
- **Teams.** Private, workspace and public visibility, roles and per-skill access, so a company has one catalog of the know-how its agents run on.

And you don't learn a CLI to use any of it. You tell your agent "Use HappySkills to install the release skill" or "…to publish my skill", and HappySkills' own skills run the commands for you. The CLI underneath was designed for agents first: JSON on every command, typed error codes and a schema the agent can read.

The honest comparison: Microsoft's APM also does dependencies and a lock file, and adds security scanning, but it installs from git with no hosted catalog, roles or merge. Vercel's skills CLI is great for grabbing public skills fast, with an experimental lock file. Anthropic's plugin marketplaces now resolve plugin dependencies, inside Claude Code only. None of them puts dependencies, a hash-checked lock, three-way merge and a team catalog together, operated by talking to your agent. That combination is why I built it.

## Use it when

- You have more than a handful of skills and they're drifting: the same skill copied into several projects, each edited a little differently.
- Skills that build on other skills: declare the dependency with a version range and one install brings the whole tree.
- A team that should share one catalog of skills, with private, workspace and public visibility, roles and per-skill access.
- You want the same agent setup on every machine: commit skills-lock.json and anyone gets exactly your versions, checked by hash.
- You use several agents (Claude Code, Codex, Cursor and others) and want one place for the skills they all read.

## Don't use it when

- You need to self-host the registry: it's a managed service, and self-hosting is a deliberate non-goal.
- You need security scanning of third-party skills before install: it doesn't do that yet. Read a public skill before you install it.
- Your skills are plugins tied to Claude Code only, with hooks and MCP servers: Anthropic's plugin marketplaces are built for that.

## Pricing

Free for individuals. Installing public skills needs no account; publishing, private skills and workspaces need a free account.

## Gotchas, with fixes

- **Gotcha:** Installs fail without a clear error on Node.js older than 22.
  **Fix:** Upgrade Node.js to 22 or later before running npx happyskills setup.
- **Gotcha:** A newly installed skill doesn't appear: agents read skills when a session starts.
  **Fix:** Restart the agent session after installing or updating skills.
- **Gotcha:** A SKILL.md over 500 lines, or a forbidden character in its frontmatter, makes the skill silently invisible to the agent.
  **Fix:** Ask "Use HappySkills to audit my skill": it flags both before you publish, and validate refuses to ship a broken one.
- **Gotcha:** Without the prefix, the agent may answer from memory instead of running HappySkills.
  **Fix:** Start requests with "Use HappySkills to …", or use the slash commands in Claude Code (/happyskills-search and the others).
- **Gotcha:** update refuses to touch a skill you've edited locally.
  **Fix:** That's on purpose: use pull, which does a three-way merge from the version you started from, or --force to discard your edits.

## Instead, consider

- [Claude Code](https://cloudlesslabs.com/stack/claude-code/index.md): The agent that reads the skills. HappySkills manages them; it doesn't replace the agent.
- [Self-hosted MCP servers](https://cloudlesslabs.com/stack/mcp-servers/index.md): Giving the agent live access to a system or API, rather than know-how and workflows.

## Sources

- [Install HappySkills](https://happyskills.ai/docs/start/install-happyskills/)
- [HappySkills FAQ](https://happyskills.ai/docs/faq/)
- [Why I Built HappySkills: Turning Expertise Into Infrastructure](https://happyskills.ai/blog/why-i-built-happyskills/)
- [Do Agent Skills Support Dependencies? Not in the Spec, So I Added Them](https://happyskills.ai/blog/agent-skills-dependencies/)
- [skills-lock.json: What It Is and How It Pins Your Skills](https://happyskills.ai/blog/the-lock-file-for-ai-agent-skills/)
- [How to Share Skills With Your Team](https://happyskills.ai/blog/share-skills-with-your-team/)
- [Agent Skills specification](https://agentskills.io/specification)

## Related

- [Build and test](https://cloudlesslabs.com/stack/layers/dev/index.md): every tool in this layer, with its verdict
- [The whole stack in one file](https://cloudlesslabs.com/stack/brief.md): every use case and tool, for questions that span many
- [Map of every Markdown file on cloudlesslabs.com, with what each one answers](https://ask.cloudlesslabs.com)