Reusable agent skills, rules, prompts, and workflows for AI-assisted development as used at Jung von Matt TECH.
Install all skills with the skills CLI — works with Claude Code, Cursor, Copilot, Codex, Cline, and more:
npx skills add jungvonmatt/agent-toolkit # install all skills
npx skills add jungvonmatt/agent-toolkit --list # browse before installingOr grab individual skills:
npx skills add jungvonmatt/agent-toolkit --skill start-ticket
npx skills add jungvonmatt/agent-toolkit --skill pr-description
npx skills add jungvonmatt/agent-toolkit --skill pr-review
npx skills add jungvonmatt/agent-toolkit --skill scheduled-pr-reviewAll skills are namespaced under jvm-skills to avoid conflicts with other skill packs.
Manual install: Clone the repo and run
./sync-skills.shto copy skills into~/.agents/skills/.
Install via the plugin marketplaces if your team does not use Node/npm locally.
claude plugin marketplace add jungvonmatt/agent-toolkit
claude plugin install jvm-skills@jvm-skillsIf your environment has no GitHub SSH keys, force HTTPS cloning:
export CLAUDE_CODE_PLUGIN_PREFER_HTTPS=1
claude plugin marketplace add jungvonmatt/agent-toolkit
claude plugin install jvm-skills@jvm-skillscodex plugin marketplace add jungvonmatt/agent-toolkit
codex plugin add jvm-skills@jvm-skills| Skill | Description |
|---|---|
write-ticket |
Turns a raw idea, bug report, or feature request into a development-ready ticket — acts as a product coach that finds weak spots, refines requirements with the author, and publishes directly to Jira, GitHub, GitLab, Linear, or Asana. |
start-ticket |
Turns a ticket (Jira, Asana, Linear, GitHub/GitLab Issues) into a ready-to-execute implementation plan — fetches the ticket, pulls design specs, explores the codebase, pressure-tests the requirement, and writes a plan. |
implement |
Executes a start-ticket plan task-by-task — builds, tests, and verifies each vertical slice, then pauses for a human to review the working-tree diff before every commit. Add auto to run the whole plan in one approved pass. |
pr-description |
Generates a concise PR/MR description in Simplified Technical English from the current branch diff — auto-captures screenshots when UI changes are detected. |
pr-review |
Comprehensive pull request review against a configurable target branch with severity-ranked findings, Jira traceability, accessibility checks, and performance analysis. |
scheduled-pr-review |
Runs one unattended first-pass review over all open PRs/MRs from a scheduled task — reviews the full PR in every round but, after round 1, posts minor findings only where the PR changed since the last round. Reviews trusted PRs in a worktree with the project's env files, uses CI results before running checks, and keeps its state per project. |
writing-tests |
Auto-invokes when authoring tests — enforces behavior-first, mutation-sensitive tests that query by ARIA role or data-testid, never by class name, and cover atomic units first. |
technical-writing |
Auto-invokes when writing comments, doc comments, documentation, or blog posts — clean-code comments that say why, one Diátaxis mode per doc page, Simplified Technical English prose, and a cut test that deletes every sentence the reader does not need. |
repo-diagnostics |
Git-based diagnostics that reveal churn hotspots, bus factor, bug clustering, commit velocity, and crisis patterns — before reading any code. |
jvm-design |
Bootstraps a project with the Jung von Matt CI 2026 design system by copying DESIGN.md, brand assets, and fonts into the project root. |
design-md |
Reverse-engineers a design system from any source (Figma, live website, screenshot, or tokens) into a spec-compliant DESIGN.md with tiered tokens. |
| Command | Description |
|---|---|
/jvm-skills:write-ticket |
Turn a raw idea into a development-ready ticket |
/jvm-skills:start-ticket |
Fetch a ticket and produce a ready-to-execute plan |
/jvm-skills:implement |
Execute a plan task-by-task, reviewing the working-tree diff before each commit |
/jvm-skills:pr-description |
Generate a PR/MR description from the current branch |
/jvm-skills:pr-review |
Review the current PR/MR for merge readiness |
/jvm-skills:scheduled-pr-review |
Run one scheduled first-pass review over all open PRs/MRs |
/jvm-skills:repo-diagnostics |
Run git-based repository diagnostics |
/jvm-skills:jvm-design |
Bootstrap the JvM CI 2026 design system into a project |
/jvm-skills:design-md |
Reverse-engineer a DESIGN.md from any design source |
Some skills route to companion skills from external sources. They degrade gracefully — they still work without them, loading only the companions present in the workspace.
# General-purpose skills (planning, spec-driven-development, security, performance …)
npx skills add addyosmani/agent-skills
# Chrome DevTools skills (a11y-debugging, network, performance traces …)
npx skills add ChromeDevTools/chrome-devtools-mcp
# Web performance skills (Core Web Vitals, loading, interaction, media …)
npx skills add nucliweb/webperf-snippets
# Alignment skills (grilling, domain modeling …)
npx skills add mattpocock/skills
start-ticketroutes toplanning-and-task-breakdown, and conditionally tospec-driven-development,doubt-driven-development,interview-me,security-and-hardening, andperformance-optimizationfromaddyosmani/agent-skills. Its stress-test step usesgrill-with-docs(fallback:grilling) frommattpocock/skillswhen present.
We don't try to cover everything ourselves. These are the skill packs and MCP servers we rely on across our projects — though not every project uses all of them. We pick what fits the stack and the task. Some of our skills reference companions from these packs but work fine without them.
| Repository | Description | Classification |
|---|---|---|
| addyosmani/agent-skills | Full-lifecycle engineering skills from spec to ship. | Engineering |
| mattpocock/skills | Alignment-first workflow with grilling, domain modeling, and TDD. | Engineering |
| obra/superpowers | Development methodology with subagent orchestration and git worktrees. | Methodology |
| GoogleChrome/modern-web-guidance | Modern web platform features and best practices by the Chrome and Edge teams. | Web Platform |
| nucliweb/webperf-snippets | DevTools snippets for measuring and debugging Core Web Vitals. | Performance |
| vercel-labs/agent-skills | React/Next.js performance, Vercel deployment, and the npx skills CLI. |
React, Vercel |
| antfu/skills | Auto-generated skills for the Vue/Vite ecosystem. | Vue, Tooling |
| Server | Repository | Description |
|---|---|---|
| Chrome DevTools | ChromeDevTools/chrome-devtools-mcp | Browser automation, performance traces, Lighthouse audits, and screenshots. |
| Atlassian | sooperset/mcp-atlassian | Jira and Confluence access for Cloud and Server/Data Center. |
| GitHub | github/github-mcp-server | Official GitHub MCP server for repos, issues, PRs, and code search. |
| Figma | GLips/Figma-Context-MCP | Figma design data and image downloads for one-shot implementation. |
| Context7 | upstash/context7 | Up-to-date, version-specific library documentation as context for coding agents. |
Contributions welcome! See Contributing Guide for guidelines.