A skill pack with seven slash commands mapping to engineering stages: spec, plan, build, test, review, simplify and ship. Demoed the interview command, which extracts requirements into a markdown spec. Compared to Garry Tan's G Stack but focused on the engineering workflow.
Agent Skills at a glance
- What is Agent Skills?
- Agent Skills is a library. MIT-licensed pack of 25 engineering skills and slash commands for AI coding agents, covering spec through ship.
- Does Agent Skills work in the EU?
- Available in the EU. MIT-licensed GitHub repository that can be cloned and installed anywhere, with no regional restrictions in the license or README. The agent it runs in determines any model-side availability. Last verified 14 Nov 2026.
- How much does Agent Skills cost?
- Open source, open source. Free under the MIT license; requires a separate AI coding agent, which may have its own costs. Licence: MIT.
- Where was Agent Skills shown?
- In 1 creator video by Matthew Berman. Most recently by Matthew Berman on 12 Jun 2026, at 8:58 in “You NEED to try these open-source AI projects RIGHT NOW”.
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
Agent Skills is an open-source collection of engineering workflow skills for AI coding agents, published by Addy Osmani on GitHub. The skills encode practices such as spec-first development, small atomic tasks, test-driven development, code review and shipping checks, so that agents follow them consistently across the development lifecycle.
The README lists 9 slash commands mapped to lifecycle stages: /spec, /plan, /build, /test, /constraints, /review, /webperf, /code-simplify and /ship. A /build auto mode generates a plan and implements every task after a single approval, with tests and per-task commits. The repo ships 25 skills in total, including interview-me (requirements interrogation), test-driven-development, code-review-and-quality, api-and-interface-design and frontend-ui-engineering. Some skills activate automatically based on the task.
It installs through the open skills CLI (npx skills add addyosmani/agent-skills), which targets 70+ agents including Claude Code, Cursor, Codex, Copilot and Cline. Native integrations are documented for Claude Code (plugin marketplace), Gemini CLI and Antigravity CLI, and Cursor setup is done by copying skills into .cursor/skills/. A noted limit: installing a single skill via npx does not copy the repo-level references/ directory, so shared checklists are missing (tracked as issue #361).
- 25 engineering skills covering define, plan, build, verify, review and ship phases
- 9 slash commands such as /spec, /plan, /build, /test, /review and /ship
- /build auto runs the approved plan task by task with tests and individual commits
- Installs into 70+ agents via the skills CLI; native plugins for Claude Code, Gemini CLI and Antigravity
- Skills activate automatically based on the work, e.g. API design or frontend UI
As seen on YouTube
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Similar AI tools
Closest neighbours by meaning, not just by category.
Matt Pocock's skills
Library · Matt Pocock
Collection of agent skills for AI coding assistants such as Codex, published by developer Matt Pocock.
GStack
Library · Garry Tan
Open-source set of Claude Code slash-command skills that act as a virtual engineering team, by Garry Tan.
Diagram Design
Library
MIT-licensed agent skill that generates brand-matched diagrams as a single self-contained HTML + SVG file.
Obsidian Skills
Library · Obsidian
Open-source agent skills that teach AI coding agents to work with Obsidian vaults, Markdown, Bases, Canvas and the CLI.
Loopy
Library · Forward Future
Open-source agent skill and web catalog of reusable, bounded AI-agent workflow "loops" with checks and stopping conditions.
Skill Editor
Agent
An agent skill that updates other skills by reviewing many recent conversations for recurring feedback.
