sharebox
Publish and share the small web tools your AI agents build, as easily as a Google Doc. Google sign-in, permissions, storage and gVisor sandboxing built in. Self-hosted.
Install / Use
claude mcp add zalaso -- npx -y github:zalaso/shareboxIf the server publishes to npm under a different name, use that package instead — check the repo README.
MCP Server
Model Context Protocol server
Quality Score
Category
Development & EngineeringSupported Platforms
Tags
Our assessment of sharebox
sharebox scores 83/100 on our quality scale, 3050th of 4,607 Development & Engineering skills we index.
Its MCP Server is 8.2 KB long, well organised into 9 sections with 7 code examples: a thorough specification that gives an agent plenty to work with.
It has 10 GitHub stars, so there is little community track record yet; judge it on its content.
Maintenance, license and trust
- The repository was last updated 2 days ago, so sharebox is actively maintained.
- It is released under AGPL-3.0, a copyleft license: you can use it, but modified versions you distribute must carry the same license.
- Its trust signals score 97/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
sharebox compared with similar skills
All 4 of these similar skills score higher than sharebox; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| sharebox (this skill)by zalaso | 83 | 10 | 2d ago | MCP Server |
| Agent-Reachby Panniantong | 100 | 96.7k | 5d ago | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 75.2k | today | CLAUDE.md |
| CowAgentby zhayujie | 100 | 47.4k | today | CLAUDE.md |
| ai-job-searchby MadsLorentzen | 100 | 46.4k | 4d ago | CLAUDE.md |
Frequently asked questions
- How do I install sharebox?
- Run
claude mcp add zalaso -- npx -y github:zalaso/sharebox. The install tabs above show the steps for each supported agent. - Which AI agents does sharebox work with?
- It is written for Claude Code and Claude Desktop, as a MCP Server file. Other agents that read the same format can often use it too.
- Is sharebox safe to use?
- It is AGPL-3.0-licensed and scores 97/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is sharebox still maintained?
- The repository was last updated 2 days ago, so sharebox is actively maintained.
Skill content
View source on GitHubShareBox
Publish and share small web tools built by AI agents — as easily as sharing a Google Doc.
English · Italiano
<img src="docs/img/demo.gif" alt="Publish with one command, share like a doc, everyone signs in with Google, revoke instantly" width="760"> </div>Anyone can now ask an AI agent for a small tool — a team vacation tracker, a form, a dashboard. Putting it online and sharing it safely is the hard part: hosting, domain, HTTPS, login, database, permissions.
ShareBox removes all of that. Your agent builds the tool and publishes it with one command; you get a link and share it with specific people, a whole company domain, or anyone with the link. Visitors sign in with Google. The tool gets their name and email for free, stores its data in its own database, and runs in its own sandbox.
"Build a vacation tracker for my team and publish it." → the agent writes it, publishes it, and replies with
https://vacation-tracker-k3x9.your-tools-domain.org→ you share it with@yourcompany.com→ colleagues open it, sign in with Google, and add their days off.
ShareBox is self-hosted: you run your own instance on a small Linux server, with your domains, your Google login and your users. It is independent from any other instance.
Features
- One-command publishing from a terminal (
sharebox publish) or straight from an AI agent through the bundled MCP server. Every tool gets its own HTTPS address; republishing updates the same tool and keeps its data. - Private by default, shared like a document: specific people (email), every Google Workspace account of a domain, or anyone with the link — each as can use or can manage. Revocations apply on the very next request.
- Google sign-in handled by the platform: tools implement no login and receive a trusted identity (name, email, role).
- Built-in storage: every tool has its own SQLite database. Static pages use a tiny SDK with server-side permissions (
sharebox.collection("vacations").add({...})); tools with server code get direct SQL access. - Strong isolation: each tool runs in its own gVisor sandbox with workerd, its own network, read-only code and CPU/memory/process limits. Only one small, separate process may talk to Docker.
- Web dashboard for sharing, status, logs, activity, suspension and deletion — plus ready-to-copy commands to connect your computer and your agents.
- Daily backups, optionally copied to Google Drive encrypted.
- Cheap: runs on a ~4 €/month VPS; the tools' domain can be a free DuckDNS subdomain.
How it works
Internet ─▶ Caddy (HTTPS) ─▶ platform: Google login, permissions, API, dashboard, gateway
│ │
│ internal network │ one internal network per tool
▼ ▼
orchestrator gVisor sandbox + workerd (one per tool)
(only one with Docker) static files, SDK, creator's worker, SQLite
Every request to a tool goes through the gateway, which checks the session and the permissions, strips anything a browser could forge and injects the visitor's identity. Tools cannot reach each other, the platform, the host or the internet. Design and rationale: docs/architettura.md and the decision records in docs/adr/ (in Italian).
Quick start
You need: an Ubuntu 24.04 server (2 GB RAM is enough to start) with ports 80/443 free, a domain for the platform (e.g. sharebox.example.com), a domain for the tools — a free DuckDNS subdomain or a domain on Cloudflare — and a Google OAuth client.
On the server, as root:
git clone https://github.com/zalaso/sharebox /opt/sharebox && cd /opt/sharebox && bash deploy/install.sh
The installer sets up Docker and gVisor, asks for your domains and credentials (secrets are not echoed), checks DNS, ports and firewall, builds everything from source, starts it and schedules nightly backups.
Then, on your computer (Node 20+), install the CLI and connect it — your browser opens to confirm:
npm install -g sharebox-cli
(Each instance also serves a copy matching its own version: npm install -g https://sharebox.example.com/cli/sharebox.tgz.)
sharebox login sharebox.example.com
And give your agent the tools (Claude Code shown; on Windows use -- cmd /c sharebox mcp):
claude mcp add --scope user sharebox -- sharebox mcp
Step-by-step guide, including DNS and Google setup: docs/install.md.
Building a tool
An agent connected through MCP reads the built-in guide (sharebox_guide) and knows what to do. By hand:
sharebox init vacation-tracker --name "Vacation tracker" # starter with an SDK example
sharebox publish vacation-tracker # → https://vacation-tracker-xxxx.<tools domain>/
sharebox share vacation-tracker @yourcompany.com # or a person's email, or "chiunque" (anyone)
A tool is a folder with public/ (HTML, CSS, JS) and, optionally, a worker.ts for server code:
<script src="/__sharebox/sdk.js"></script>
<script>
const me = await sharebox.me(); // { email, name, role }
const days = sharebox.collection("vacations");
await days.add({ from: "2026-08-01", to: "2026-08-15" });
const all = await days.list(); // records with owner and canEdit
</script>
Users with can use edit only their own records; can manage edits everything — enforced on the server. Full CLI reference: docs/cli.md.
Complete example: examples/ferie — the team vacation tracker from the demo, built by an agent through MCP: team calendar, upcoming absences with working-day counts (Italian public holidays included), yearly summary. Plain HTML/CSS/JS, no dependencies.
Status
Young project, running on the author's private instance. Working today: publishing, sharing, storage, CLI, MCP, dashboard, backups, self-hosted installer, CI.
Known limitations and next steps:
- The interface (dashboard, pages, CLI, installer) is in English and Italian, picked from the browser or system language. Other languages are welcome: the dictionaries are in
packages/platform/src/messages.ts,packages/platform/dashboard/app.js,packages/runtime/src/messages.tsandpackages/cli/src/i18n.ts. - Single server: about 25 tools running at once on 2 GB of RAM; stopping idle tools is planned.
- Tools cannot call external APIs yet (no internet access, no secrets) — planned, behind an egress proxy.
- Sign-in with Google only; publishing is limited to the e-mails listed in
CREATOR_EMAILS.
Documentation
| | | |---|---| | Install your own instance · Installazione | Requirements, DNS, Google, installer | | Running an instance | Updates, administration, checks, backups (Italian) | | Restore from backup | (Italian) | | CLI and MCP | (Italian) | | Architecture and decisions | (Italian) |
Contributing
Bug reports, ideas and pull requests are welcome — see CONTRIBUTING.md. Please report security issues privately: SECURITY.md.
License
GNU AGPL v3 or later. You may use, study, modify and redistribute ShareBox, including to run a service; if you run a service on a modified version, you must make your changes available to its users.
Related Skills
Agent-Reach
96.7kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
headroom
75.2kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
CowAgent
47.4kOpen-source personal AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-agent, multi-model, multi-channel. Lightweight, extensible, one-line install.
ai-job-search
46.4kThe job search that runs on your machine. AI job application framework built on Claude Code: evaluate postings, tailor CVs, write cover letters, prep interviews. Fork it and own it.
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
