Public Alphav0.1

Skillmine

The native lifecycle for
AI assistant skills

Create, register, install, sync, and diagnose skills across Claude Code and OpenCode. Built in Rust. Powered by Git.

$init
$create
$add
$install
$sync
$doctor
skillmine
$skillmine create my-skill
Created ./my-skill with SKILL.toml and starter files
$skillmine add ./my-skill
Registered my-skill in managed configuration
$skillmine install
Installing 3 skills...
✓ Hello-World
✓ git-release
✓ python-testing
$skillmine sync --target=claude
Synced 3 skills to ~/.claude/skills/ (public alpha target)
$skillmine doctor
✓ Configuration valid
✓ Lifecycle healthy
✓ No drift detected
$
12+
Commands
2
Targets
Rust
Language

Managing AI skills is a mess

You have powerful AI assistants, but no good way to create, organize, version, and sync the skills that make them useful.

Skill Chaos

AI skills scattered across different assistants, versions, and locations. No single source of truth for your workflow enhancements.

Version Drift

Skills get updated silently. Your carefully tuned prompts break overnight. No way to lock to known-good versions or track changes.

Manual Sync Hell

Copying files between Claude Code and OpenCode. Updating one assistant still means repeating runtime setup by hand.

No Visibility

No way to see what skills are installed, which are outdated, or diagnose why something broke. Blind debugging.

The closed-loop lifecycle

Create, add or register, install, sync, and doctor. One continuous workflow from authoring to runtime health.

01

Create

Generate a valid local skill package skeleton so the lifecycle starts and stays in one native tool.

$skillmine create my-skill
02

Add / Register

Register the generated local skill or a remote source in declarative config as managed state.

$skillmine add ./my-skill
03

Install

Resolve and cache skills in content-addressable storage. Deterministic, reproducible, and reviewable.

$skillmine install
04

Sync

Expose skills to your assistant runtime. Public alpha supports Claude Code and OpenCode.

$skillmine sync --target=claude
05

Doctor

Run diagnostics to check configuration, detect drift, and validate the whole loop end to end.

$skillmine doctor
1
2
3
4
5
Complete lifecycle in one tool

Built for AI-native workflows

Not just downstream install and sync. A public alpha lifecycle for creating, managing, shipping, and validating AI skills.

Native Skill Creation

Start the lifecycle with `skillmine create` and keep creation, registration, install, sync, and diagnostics in one tool.

C
O
Supported targets

Declarative Configuration

Define your desired state in skills.toml. Git-trackable, reviewable, shareable.

Deterministic State

Content-addressable storage ensures identical inputs produce identical outputs. Reproducible everywhere.

Alpha Runtime Support

Sync skills to Claude Code and OpenCode in the current public alpha. One config, supported targets kept explicit.

Version Locking

Lock files pin exact versions. Update when you choose, not when upstream changes.

Drift Detection

skillmine doctor validates state and detects drift. Know when things change.

Git-Native Workflow

Skills are just Git repos. Branch, tag, fork, and PR your AI workflows like code.

Ready to mine
some skills?

Install Skillmine and run the full loop from create to doctor with one CLI.

Public Alpha Status
Install via Cargo
$cargo install skillmine
Supported Runtime Targets
Claude Code and OpenCode
Known Limitations
No Cursor target in this alpha. Website is documentation-only.