Last verified
Sync AI coding skills, agents, MCP connections, and hooks across Claude Code, Codex, and Pi from one Git-backed source.
Skillshare maintains one set of skills and syncs them to every AI tool in the workflow. Refining a code review skill propagates to all configured targets with one sync command, with no duplication or drift between tools.
You use exactly one AI coding tool and have no plans to add others. A dedicated manager adds overhead not justified by a single-target workflow.
Project skills in .skillshare/ ship with the repo. A new teammate clones the repo, runs skillshare sync, and immediately has the team's code review checklist and context files in every AI tool they use. No copy-pasting from wikis or chat threads.
Your team standardizes on one AI tool with built-in team settings that already handle skill distribution. A separate manager adds a second sync layer to maintain.
Version the AI coding setup in Git alongside dotfiles. On a new machine, clone the config repo and run skillshare sync to restore skills, agents, and MCP connections to every configured tool. The backup command creates a snapshot before making large changes.
You rarely switch machines and do not manage AI tool configuration in version control. The benefit of Git-backed sync is highest across multiple machines.
A .skillshare/ directory in the repo distributes contribution guidelines, code style skills, and testing checklists to contributors. Contributors install declared skills with one command and get the same AI context the maintainer uses, reducing onboarding friction.
Your contributors use AI tools that do not support skill files. Skillshare targets tools with SKILL.md-style configuration; tools with proprietary formats cannot be targeted.
Managing AI coding tool setups is a copy-paste problem. Each tool stores its skills, system prompts, and configuration separately. When you switch from Claude Code to Codex, or add Pi to your workflow, you rebuild your setup from scratch. Teammates copy instructions from chat logs or Slack threads instead of installing a shared checklist.
The problem compounds on new machines. There is no standard way to version your AI coding setup alongside your code, so setup steps fall through the gaps of your dotfiles or onboarding docs. Teams running multiple AI tools end up with each developer carrying a slightly different set of skills, which produces inconsistent code review, testing, and formatting results across the team.
A source directory at ~/.config/skillshare/ distributes skills to Claude Code, Codex, Pi, OpenCode, and other tools as symlinks on macOS/Linux, with NTFS junctions on Windows requiring no admin rights. One skillshare sync command pushes the source to every configured target, so changes propagate everywhere without touching each tool individually.
Every skill installation runs through an audit gate before the files reach your AI tool. Critical audit findings block the install automatically. Running skillshare audit on demand scans existing skills for issues. This differs from a manual copy, where no verification happens before content reaches the coding agent.
Initializing a project with skillshare init -p creates a .skillshare/ directory in the repo. These skills commit with the code and scope to the project, so teammates clone the repo and sync their AI tools without a separate setup step. Project skills do not affect the global source directory.
Define MCP server connections once in config.yaml and sync them to Claude Code, Codex, Pi, VS Code, OpenCode, and other MCP-compatible tools with skillshare sync mcp. A dry-run flag previews native configuration changes before applying them, and environment variable references keep credentials out of the config file.
Push the source to any Git host (GitHub, GitLab, or self-hosted) with skillshare push and pull it on another machine with skillshare pull. Tracked repos let a team maintain a shared skill library in a separate Git repo; teammates install it once and update with skillshare update --all.
Running skillshare ui opens a web dashboard for managing skills, agents, MCP, and hooks without the CLI. The companion desktop app (macOS Apple Silicon, Windows, Linux) adds a native window with guided first-launch setup. Install the desktop app on macOS with brew install --cask skillshare-app.
Most developers treat AI coding tool configuration like dotfiles: each tool gets set up once per machine, there is no shared source, and new machines start from scratch. This imperative approach works for a single developer with one tool, but does not scale to a team or multiple tools.
| Feature | Skillshare | Manual per-tool install |
|---|---|---|
| License | MIT | n/a |
| Source of truth | Single directory, Git-backed | Separate per tool |
| New machine setup | git clone config + skillshare sync | Re-run each install manually |
| Security audit | Built-in; critical findings block install | None |
| Team skills | Tracked repos, project .skillshare/ | Copy from chat or docs |
| Runtime requirement | None (Go binary) | Varies by tool |
Skillshare is the better choice when you use more than one AI coding tool, manage a team, or switch machines frequently. The declarative source model means your AI setup travels with your dotfiles, and project skills live alongside the code.
The manual approach is still reasonable for a single developer using one AI tool with no plans to switch tools or machines.
Paid team tiers from AI coding platforms like Cursor Teams and GitHub Copilot Enterprise offer centralized configuration, organization-level policy management, and usage analytics as proprietary services. They are designed for a single vendor's tooling.
Skillshare works across vendors: Claude Code, Codex, Pi, and others share the same source directory. There is no per-seat fee, no cloud component, and no vendor dependency. Teams that run multiple AI tools, or want to avoid locking team configuration into one vendor's paid plan, get more flexibility from skillshare.
Cursor Teams or Copilot Enterprise are worth considering when centralized policy enforcement, compliance logging, or usage analytics per seat are requirements. Skillshare does not provide those controls.
Install the CLI on macOS with Homebrew, or run the macOS and Linux install script.
```bash
brew install skillshare
```
The macOS and Linux install script handles setup without requiring Homebrew.
```bash
curl -fsSL https://raw.githubusercontent.com/runkids/skillshare/main/install.sh | sh
```
Windows PowerShell setup uses the official install script.
```powershell
irm https://raw.githubusercontent.com/runkids/skillshare/main/install.ps1 | iex
```Skillshare supports Claude Code, Codex, Pi, OpenCode, and OpenClaw, plus any tool that uses SKILL.md-style configuration files. Run skillshare status to see which tools are detected on your current machine. New tools can be added as they adopt compatible skill file formats.
Yes. Skillshare is MIT licensed and free to run, including for commercial use. The CLI, desktop app, and all features are open source with no paid tier. The source code is at github.com/runkids/skillshare.
No. Skillshare runs entirely on your own machine. The source directory lives at ~/.config/skillshare/ and skills sync to AI tools via symlinks or copies on the same machine. Remote sync goes to your own Git host (GitHub, GitLab, or self-hosted), not to any skillshare-operated service.
Merge mode, the default, never overwrites a skill a tool already has. Skillshare adds what is in the source that the tool lacks and leaves existing tool-managed skills in place. Use skillshare diff to preview what would change before running a full sync.
Yes. The runkids/setup-skillshare@v1 GitHub Action installs the CLI and accepts a source path, with options for audit, project mode, and version pinning. Add the action as a step, then run skillshare sync to make project skills available to AI tools running in the CI environment.
Alternative to
Grok Bot
Self-hosted AI agent: persistent memory, multi-channel chat, model choice.
Alternative to
LM Studio
Local LLM server for Apple Silicon, faster than LM Studio
Local LLM inference for Apple Silicon and NVIDIA GPUs, no Python.
Desktop computer use for AI agents via OS accessibility tree
Persistent local memory and rules for Claude Code, Codex and Cursor
Official Claude API notebooks and code recipes for Python developers
Sponsors
Alternative to
Superhuman
AI email assistant that drafts replies and clears your inbox
Alternative toGrok Bot
AI teammates you own: your keys, your model, your machine.
Live monitoring on LinkedIn, X, Reddit & more. Agentic social listening for startups.
The projects we add and every new guide. Free, unsubscribe any time.
Claude Code to 
OpenCode
Jira to 
Windshift
Notion to 
AppFlowy
Slack to 
Mattermost
Linear to 
GitLab
Wispr Flow to 
Handy
Arc to 
Zen Browser
CapCut to 
Concat
Grok Bot to 
rakazo
Superhuman to 
Inbox Zero
Airtable to 
AITable
Muse to 
OpenClaw
Railway to 
Coolify
GitHub Copilot to 
Cline
Datadog to 
Netdata
Microsoft Teams to 
Rocket.ChatConfluence to 
Docmost
Pinecone to 
Qdrant
Zapier to 
n8n
Devin AI to 
OpenHands
Firebase to 
Supabase
1Password to 
KeePassXC
Contentful to 
Strapi
Adobe Premiere Pro to 
Shotcut
