Software engineer Sultan Valiyev has built a solution to a problem that didn’t exist two years ago: how to manage the dozens or hundreds of instruction files that teach AI coding assistants how to work with your specific tools and workflows.
As developers increasingly rely on AI assistants like Claude Code, Cursor, and GitHub Copilot to help write code, they’ve started accumulating “skills” — specialized markdown files that give these agents context about codebases, frameworks, and development environments. But without a centralized system, these files pile up across configuration directories with no easy way to track what’s installed where.
That’s where SkillsGate’s skill management tool comes in. The free, open source desktop app and terminal interface gives developers a single place to browse, install, edit, and organize AI agent skills across 19 different coding assistants.
Built Out of Personal Frustration
Valiyev, based in California, created the entire system as a hobby project using TypeScript, React, and SQLite. The motivation was straightforward: he was personally dealing with the mess of managing agent skills across multiple AI tools and wanted something better.
The project has found an audience among developers facing the same headaches. The open source repository has earned 250 stars on GitHub, while the command-line and terminal UI packages have been downloaded more than 3,300 times combined through npm. What started as a weekend side project to scratch his own itch has grown into a tool other developers find useful too
Local-First, No Strings Attached
Everything in SkillsGate’s developer workflow tools runs locally. There’s no account creation, no sign-up flow, and no cloud dependency. Skills live on the user’s filesystem, and the app reads and writes them directly. Settings sync between the desktop app and terminal UI through a shared SQLite database stored on the machine.
The app connects to the skills.sh catalog of more than 91,000 public skills for discovery, but adds a crucial management layer on top: per-agent filtering, one-click installation to specific assistants, and the ability to remove a skill from one tool without affecting others. The built-in editor makes the editing experience familiar to most developers.
The terminal UI can be installed with a single command: npx skillsgate. The desktop app is available as a download from GitHub Releases for macOS, Windows, and Linux
What Comes Next
Valiyev hopes the project eventually makes AI coding skill management something developers don’t have to think about. Near-term plans include adding collections that let users group skills by project or workflow and apply them to agents in one click, drag-and-drop organization, and local favorites for quick access.
The focus remains on local-first, open source tooling. For developers tired of hunting through config files to figure out which AI assistant knows what, that local-first, no-setup approach resonates strongly.