Sx 2.0 - Share AI Skills Through Dropbox or Google Drive
Sx 2.0 turns any shared folder into a skill server for AI agents. No server, no git, no account needed - just point Dropbox at your team vault and share skills instantly.
TL;DR
TL;DR: Sx 2.0 lets you share AI skills with your whole team by dropping them in a Dropbox or Google Drive folder - no server, git knowledge, or account required.
Source and Accuracy Notes
- Project page: sleuth.io (links to app at skills.new)
- Source repository: github.com/sleuth-io/sx
- License: Apache-2.0 (verified via
LICENSEfile on main branch) - HN launch thread: news.ycombinator.com/item?id=48900319 (44 points, 2026-07-13)
- GitHub stars: 296 (verified via GitHub REST API)
- Source last checked: 2026-08-20
What Is Sx?
Sx (pronounced “ess-ex”) is an open-source tool that turns any folder your team already syncs - Dropbox, Google Drive, iCloud, OneDrive - into a shared vault of AI skills. Skills live in a manifest file inside that folder, and every team member’s AI tools pick them up automatically.
The core problem it solves: AI assets (skills, MCPs, prompts, rules) usually live inside one Git repo. Sharing them across repos means copy-pasting, and they drift out of sync with no source of truth. Sx gives you a real sharing layer without requiring git or any infrastructure.
Sx 2.0 added a GUI app on top of the CLI, so non-technical teammates can share skills without touching a terminal. Both the app and CLI share the same underlying library - a skill added in the app appears in the CLI immediately.
Key features:
- Zero infrastructure - a shared folder is the vault; no server, no accounts
- No git required - non-technical users can participate
- Works with every major AI client - Claude Code, Copilot, Cursor, Gemini CLI, Codex, and more
- Scoped distribution - control which orgs, repos, teams, or users get which skills
- Versioning and audit trail - built-in
sx statsandsx auditcommands - Self-hostable - the vault is just a folder; you own the data
Setup Workflow
Step 1: Install the CLI
macOS/Linux via Homebrew:
brew tap sleuth-io/tap
brew install sx
Via shell script (any OS):
curl -fsSL https://raw.githubusercontent.com/sleuth-io/sx/main/install.sh | bash
Step 2: Install the GUI App (Optional)
Download from github.com/sleuth-io/sx/releases - look for the sx-app-* artifacts for macOS, Windows, or Linux. The app ships a bundled CLI inside itself, so installing the app also gives you the command-line tool.
Step 3: Point at a Shared Folder
sx init --type path --path ~/Dropbox/sx-vault
Replace the path with your team’s existing shared folder (Google Drive, iCloud, OneDrive all work). The folder becomes the team vault.
Step 4: Add and Share a Skill
# Add a skill from your local AI tools
sx add ~/.claude/skills/my-skill
# Publish it to the shared vault
sx publish
Teammates who have initialized their vault with the same folder path get the skill automatically in every AI tool they use.
Underlying Architecture
The vault is a folder containing two files:
sx.json- the manifest listing available skillssx.lock- a lockfile pinning exact versions (auto-managed)
When you run sx add, the CLI copies the skill into the vault folder and updates the manifest. When teammates sync, their local AI clients see the updated manifest and pull the new assets.
For AI clients that use git-based hooks (GitHub Copilot’s .github/hooks/sx.json, Kiro’s .kiro/hooks/), the CLI resolves from PATH so those hooks work across machines without being tied to one installation.
Supported AI Clients
From the README and FAQ, Sx integrates with:
- Claude Code
- GitHub Copilot
- Cursor
- Gemini CLI
- Codex (OpenAI)
- Cline
- OpenClaw
- Any MCP-compatible client
The distribution layer works by writing standardized hook files that each AI client reads on startup.
Limitations and Considerations
- No built-in access control - any teammate with the folder can read all skills. For sensitive prompts or credentials, encrypt the vault folder separately (BoxCryptor, Cryptomator, etc.) before syncing.
- Folder-based scoping - teams needing fine-grained per-skill permissions need a different solution; Sx scopes to org/repo/team via configuration but relies on the folder being the trust boundary.
- CLI updates - the bundled CLI inside the app does not self-update (to avoid breaking signed app bundles). Install the standalone CLI separately if you want auto-updates.
- No hosted option - if you need a hosted solution with SSO and audit logs, Sleuth offers that as a paid product at sleuth.io. The open-source Sx covers the self-hosted case.
FAQ
Q: How is this different from just sharing a Git repo of skills? A: Git requires technical knowledge to use correctly, and a skill committed to one repo is invisible to agents in other repos. Sx’s manifest system makes skills discoverable across every AI client on every machine without requiring git or repository-level configuration.
Q: Does this work with web-based AI tools? A: Yes - the README explicitly states “every AI tool, including the web ones.” The client installs a local hook that intercepts skill requests from any connected AI tool, not just terminal-based ones.
Q: Can I use Sx without any cloud storage? A: Yes. Any local folder works, including network-mounted drives or rsync-based setups. The cloud storage option (Dropbox, Google Drive, etc.) is purely for syncing the vault across machines without running your own server.
Q: What happens when two teammates edit the same skill? A: Sx uses standard file syncing from your storage provider. Conflicts are resolved by the underlying sync tool (Dropbox keeps version history and lets you restore previous versions).
Q: Is there a hosted version with more features? A: Yes, Sleuth offers a paid SaaS product with SSO, audit logs, and RBAC at sleuth.io. The open-source Sx is the self-hosted alternative.
Conclusion
Sx 2.0 solves the AI asset distribution problem cleanly: instead of building yet another server or asking everyone to learn git, it turns the folder your team already shares into a skill distribution hub. The GUI app lowers the barrier for non-technical teammates, and the Apache-2.0 license means you can self-host without vendor lock-in.
If your team uses multiple AI coding tools and struggles to keep skills in sync across them, Sx is worth a look. The 30-second demo on the project page is the quickest way to see if it fits your workflow.
Links:
- github.com/sleuth-io/sx - releases, docs, source
- sleuth.io - managed product with SSO and audit logs
Related Posts
ai-setup
Recall – Persistent Memory for Claude Code via MCP Hooks
Recall gives Claude Code a permanent memory store that survives session restarts and context compaction. Four hooks capture and restore context automatically — with cloud SaaS or self-hosted options.
2/28/2026
dev-tools
Automotive Skills Suite for AI Engineering
Evaluate Automotive Skills Suite for APQP, ASPICE, HARA, safety-plan, and DIA workflows with setup notes, governance risks, and SME review guidance.
5/28/2026
dev-tools
awesome-agentic-ai-zh Roadmap Guide
Explore awesome-agentic-ai-zh as a Chinese agentic AI learning roadmap, with setup notes, track selection, study workflow, and evaluation guidance.
5/28/2026