- Home
- Skills
- Agents & Orchestration
- Claude Code Plugin Builder: Skills, Hooks & MCP to Plugin
Works with the AI Agents you already use
Claude Code Plugin Builder: Skills, Hooks & MCP to Plugin
A loose `.claude/` folder only helps the person who made it.
$14
Claude Code Plugin Builder: Skills, Hooks & MCP to Plugin
Example session with this skill installed
Turn my .claude folder into a plugin I can share with my team. I have 4 skills, 3 slash commands, a code-reviewer subagent, a format-on-save hook and a Postgres MCP server.
- Read your context and instructions
- Compiled the claude code plugin
PLUGIN: team-dev-kit v1.0.0 · VERDICT: FIX FIRST → SHIP after 3 fixes
Components: 4 skills · 3 commands · 1 agent · 1 hook · 1 MCP server
What I built
team-dev-kit/
├── .claude-plugin/plugin.json
├── skills/ (4)
├── commands/ (review.md, ship.md, standup.md)
├── agents/code-reviewer.md
├── hooks/hooks.json
├── scripts/format.sh
├── .mcp.json
└── README.md, CHANGELOG.md, LICENSE
Marketplace repo: acme-claude-plugins/.claude-plugin/marketplace.json
Fixes applied
| Issue | Found | Fix |
|---|---|---|
| Hook used a path on your laptop | /Users/sam/.claude/format.sh | Now runs from the plugin root variable; script made executable |
| Database password in MCP config | .mcp.json line 7 | Replaced with an environment variable; documented in README |
| Hook matched every tool | "matcher": "*" | Narrowed to Write and Edit |
Verification (clean test project)
- ✓ 4/4 skills trigger on their descriptions
- ✓ /review, /ship, /standup listed
- ✓ code-reviewer agent available
- ✓ format hook fires on a test edit
- ✓ Postgres MCP tools appear when the env var is set
- ✓ Uninstall leaves nothing behind
Install (send this to your team)
/plugin marketplace add acme/acme-claude-plugins
/plugin install team-dev-kit@acme-claude-plugins
Connects securely to your tools. The creator never sees your data.
About this skill
A loose .claude/ folder only helps the person who made it. Plugins bundle skills, commands, agents, hooks and MCP servers so a teammate, client or customer can install, update and remove them with two commands. They're also where most first attempts break: component folders in the wrong place, hooks pointing at paths on your laptop, secrets left in .mcp.json.
This skill builds the package correctly the first time.
What it does:
- Checks the current plugin spec for your installed Claude Code version before scaffolding, and follows the docs wherever they've changed
- Inventories your skills, commands, agents, hooks and MCP servers and asks which to include, leaving out anything with secrets or machine-specific config
- Scaffolds the right layout:
.claude-plugin/plugin.jsonmanifest, component folders at the plugin root, scripts, README, CHANGELOG, LICENSE - Fixes paths, the #1 cause of broken plugins: rewrites hook and MCP commands to use the plugin root variable, sets shebangs and executable bits, swaps secrets for documented environment variables
- Safety-reviews every hook (narrow matchers, quoted variables, no hidden network calls) and every MCP server (pinned versions, least-privilege scopes)
- Builds a marketplace repo so people install with
/plugin marketplace addand/plugin install, including how to share it with a whole team - Validates and tests: runs the CLI validator where available, installs in a clean project, confirms each skill, command, agent, hook and MCP tool actually loads, then uninstalls cleanly
You get: a ready-to-commit plugin folder (and marketplace folder), a README with install, requirements and uninstall steps, starter templates for plugin.json, marketplace.json and hooks.json, and a verification report with a SHIP / FIX FIRST / BLOCK verdict.
Ideal if you sell or share Claude Code setups, run an internal tooling team, or want your skills discoverable as a plugin.
Builds plugins for Claude Code only. Plugin field names and commands change between Claude Code versions, so it checks the current docs first and follows them. Testing the MCP server needs any credentials it uses to be available as environment variables.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 19 days ago