Skillshare turns one source directory into the source of truth for skills used by Codex, Claude, Cursor, and other AI tools. This guide covers installation, first-time setup, safe migration, synchronization, and multi-machine use.

What problem Skillshare solves

AI coding tools usually keep skills in different directories. Copying the same skill into each directory creates three recurring problems:

  • edits drift between tools;
  • it becomes unclear which copy is authoritative;
  • moving to a second computer requires another manual installation pass.

Skillshare replaces those copies with a one-way model:

source directory
    ├── skill-a/
    ├── skill-b/
    └── skill-c/
          ↓ skillshare sync
Claude / Codex / Cursor / other configured targets

The source directory remains authoritative. Targets receive symlinks, junctions, or copied files according to their configured sync mode.

This model supports three common workflows:

  • One computer: edit once, then run skillshare sync.
  • Multiple computers: store the source in Git and use skillshare push and skillshare pull.
  • Third-party skills: install through Skillshare and inspect them with its built-in security audit.

Install and update the CLI

macOS and Linux

Install with the official script:

curl -fsSL https://raw.githubusercontent.com/runkids/skillshare/main/install.sh | sh

Homebrew is also supported:

brew install skillshare

The Homebrew package can lag behind the direct installer. Use the installed CLI’s help output when a flag differs from this guide.

Windows PowerShell

irm https://raw.githubusercontent.com/runkids/skillshare/main/install.ps1 | iex

Upgrade

skillshare upgrade

During this article’s verification, the local CLI was 0.20.21 and reported 0.20.22 as available. Skillshare evolves quickly, so treat skillshare <command> --help as the command-level source of truth for your installed version.

Initialize the source and targets

For an interactive setup, run:

skillshare init

The initialization flow creates the configuration and source directories, detects installed AI tools, and asks which targets to manage. On macOS and Linux, the default global source is:

~/.config/skillshare/skills/

After initialization, inspect the result:

skillshare status
skillshare target list

Do not skip this check on an existing machine. A target list is operational configuration: it determines which directories a later sync may change.

For unattended environments, pass explicit flags instead of relying on prompts. A typical global setup is:

skillshare init \
  --all-targets \
  --no-copy \
  --no-skill

Choose targets more narrowly in CI, devcontainers, or shared environments.

Migrate existing skills without reversing the source of truth

If you already maintain skills inside Claude, Codex, or another tool, you may want to collect them into Skillshare once. Preview the migration first:

skillshare collect --all --dry-run

Only remove --dry-run after confirming that the discovered files genuinely belong in the shared source.

Avoid collecting:

  • built-in skills shipped by a tool;
  • caches or generated files;
  • target-local experiments that should remain private to one tool;
  • stale copies of skills already present in the source.

collect is a migration and recovery operation, not the normal direction of daily synchronization. After setup, the intended flow is source to targets.

Install or create skills

Install from a Git repository

skillshare install github.com/<owner>/<repository>
skillshare sync

Skillshare runs a security audit during installation. Critical findings can block the operation. Review findings before deciding whether an override is justified.

You can also audit the current source directly:

skillshare audit

Create your own skill

skillshare new my-skill

The command scaffolds a skill directory and SKILL.md. Edit the new skill in the source directory, then synchronize it:

skillshare sync

Understand sync modes

Run synchronization after adding, updating, removing, enabling, disabling, or retargeting skills:

skillshare sync

Current Skillshare documentation distinguishes three modes:

Mode Behavior Suitable for
merge Creates per-skill links while preserving target-local skills Most personal setups
symlink Replaces the whole target directory with one link Fully managed targets
copy Writes real files on each sync Containers, restricted filesystems, or tools that cannot follow links

On Windows, linked targets can use NTFS junctions. If a target cannot use links, switch only that target to copy mode:

skillshare target claude --mode copy
skillshare sync

Copy mode introduces a delay: source edits do not reach the target until the next sync. It also makes target-local edits easier to lose, which is another reason to keep the source authoritative.

Commands worth remembering

Command Purpose
skillshare status Inspect the source, targets, sync state, audit policy, and version
skillshare list List discovered skills
skillshare target list Inspect configured targets
skillshare sync Apply the source to targets
skillshare check Check tracked skills for updates
skillshare update --all Update all tracked skill repositories
skillshare uninstall <name> Move a skill to Skillshare trash
skillshare enable <name> Restore a disabled skill to synchronization
skillshare disable <name> Exclude a skill from synchronization
skillshare audit Run the security audit
skillshare doctor Diagnose configuration and target problems
skillshare ui Open the local web dashboard

After a mutation, run skillshare sync unless the command documentation explicitly states otherwise.

Synchronize across computers

Skillshare can bind the source to a Git remote. Create an empty repository on GitHub, Gitee, or another Git host, then initialize with its URL:

skillshare init --remote <repository-url>

On the first computer, publish an intentional checkpoint:

skillshare push -m "Update shared skills"

On another computer:

skillshare pull

pull updates the source and then synchronizes configured targets.

If the local source contains uncommitted work, preserve it before pulling. One option is to checkpoint it:

skillshare push -m "Save local skill changes"
skillshare pull

For a manual Git workflow:

cd ~/.config/skillshare/skills
git stash
skillshare pull
git stash pop

Resolve real merge conflicts in the source repository, commit the resolution, and then run:

skillshare sync

A safe operating routine

For a stable setup, keep the routine small:

  1. Treat the Skillshare source as the only editable copy.
  2. Run skillshare status before changing targets or migrating existing files.
  3. Preview collection and broad synchronization changes with --dry-run where supported.
  4. Audit third-party skills before trusting them with tools or data.
  5. Run skillshare sync after source or target changes.
  6. Use Git checkpoints before risky migrations or cross-machine merges.
  7. Use skillshare doctor when targets disappear or links become inconsistent.

The essential idea is not the command list. It is maintaining a clear direction of ownership: one reviewed source, many generated targets.

Official documentation: Skillshare docs