M.D.N Tech
M.D.N Tech

Open source · MIT · Free forever

Make Claude Coderemember your project.

One free skill, four commands — /handoff start, wrap, save, docs — that turn every Claude Code session into the next one's starting point. No vendor lock-in. Install in one line.

MIT licensed · 0 dependencies · Works on macOS / Linux / Windows

What it solves

Three pains. One workflow.

"I waste 10 minutes re-explaining context every session."

/handoff start

Reads your handoff file and briefs Claude on the repo state in one command.

"I lose track of what I tried and why."

/handoff wrap

Writes a session summary, rotates old history to an archive, and commits locally — every time. Documentation stays current automatically.

"Docs go stale the moment I move on."

/handoff docs

Refreshes handoff and project docs from the latest session work — no manual writing required.

Install in 30 seconds

One command. One skill.

Paste into your terminal. We auto-detected your shell — switch tabs if we got it wrong.

bash / zsh
git clone https://github.com/bugiiiii11/handoff.git && \
  mkdir -p ~/.claude/skills && \
  cp -r handoff/skills/* ~/.claude/skills/ && \
  echo "✓ Installed. Run /handoff start in Claude Code to verify."
✓

Verify: open Claude Code in any project and type /handoff start. You'll see the new skill in the menu.

Prefer to inspect first? Install manually →

1. Clone the repo:

git clone https://github.com/bugiiiii11/handoff.git

2. Read the single SKILL.md file at handoff/skills/handoff/SKILL.md.

3. Copy the skill folder into ~/.claude/skills/ (or $HOME\.claude\skills\ on Windows).

Optional: enable auto-wrap hooks (real context-usage nudges) →

Two small bash scripts read the real context size from the session transcript. At 20% of the window they ask for a /handoff docs checkpoint and you keep working; at 22% they ask for the full /handoff wrap — early enough that state gets written before long context starts blurring the answers. Requires jq and bash (Git Bash on Windows works).

1. Copy the hook scripts into your project:

mkdir -p .claude/hooks && cp handoff/hooks/*.sh .claude/hooks/

2. Register both hooks in your project's .claude/settings.json:

{
  "hooks": {
    "Stop": [{ "hooks": [{ "type": "command",
      "command": "bash \"$CLAUDE_PROJECT_DIR/.claude/hooks/auto-wrap.sh\"", "timeout": 10 }] }],
    "UserPromptSubmit": [{ "hooks": [{ "type": "command",
      "command": "bash \"$CLAUDE_PROJECT_DIR/.claude/hooks/context-warn.sh\"", "timeout": 10 }] }]
  }
}

3. Restart Claude Code — hooks load at session start. Tune with AUTOWRAP_WINDOW / AUTOWRAP_SOFT_PCT / AUTOWRAP_HARD_PCT.

Uninstall in one line: rm -rf ~/.claude/skills/handoff

One skill. Four commands.

Small surface. Sharp edges.

/handoff start

View source

Session briefing — reads the capped handoff.md, checks git status and unpushed commits, consumes any emergency snapshot. Read-only.

Use it when: At the beginning of every session — Claude needs the briefing before suggesting next steps.

> /handoff start

Session Briefing
Last session: 44 — Handoff v3...
Unpushed: 2 · Last commit: bc74d79

/handoff wrap

View source

Updates handoff.md (docs before commit), rotates overflow to handoff-archive.md, commits locally. Never pushes unless you ask.

Use it when: Mid-session checkpoint or end-of-session cleanup — idempotent, safe to call anytime.

> /handoff wrap

handoff.md → Session 45 section
Rotated 1 section to archive
✓ Committed locally (no push)

/handoff save

View source

Emergency context save — dumps session state to emergency-snapshot.md in a handful of tool calls. No commit, no questions.

Use it when: When context is nearly full and you need to preserve what happened this session, fast.

> /handoff save

Writing emergency-snapshot.md...
✓ Saved. /handoff start next session resumes.

/handoff docs

View source

Docs-only refresh — runs the wrap’s handoff.md update and archive rotation without committing.

Use it when: After a focused chunk of work, when you want the docs to reflect reality but aren’t ready to commit.

> /handoff docs

handoff.md → updating Session 45 entry
No commit — docs only

Curated skills

Skills we use in production.

Battle-tested skills from Anthropic and the broader Claude Code ecosystem. Each has shipped real work for us — no curation theater.

Marketing

Build KB

Scan your project and generate a chatbot-ready knowledge-base.md in one shot.

  • Chatbot knowledge base
  • Content audit
  • ChatKit onboarding
By M.D.N TechSource

Development

Polish

Audit any surface, rank the best wins by value over effort, then ship them one sprint at a time.

  • Performance and feel passes
  • Ranking a backlog by value over effort
  • Shipping improvements in verifiable sprints
By M.D.N TechSource

Design

UI/UX Pro Max

50+ design styles, 161 color palettes, component specs. Production-grade UI design.

  • Component design
  • Design systems
  • UI/UX specification
By nextlevelbuilderSource

SEO

SEO Audit

Full website SEO analysis. Crawls pages, detects issues, generates action plan.

  • Website optimization
  • Competitive analysis
  • Ranking improvement
By AgriciDanielSource

Marketing

Marketing Skills

Marketing skill collection for AI agents — SEO, CRO, copywriting, paid ads, growth.

  • Conversion optimization
  • Copywriting
  • Growth engineering
By Corey HainesSource

Design

Frontend Design

Build distinctive web components & pages. Production-grade UI quality.

  • Web component building
  • Page layout design
  • UI implementation
By AnthropicSource

Infrastructure

Claude API

Build Claude API applications. Prompt caching, tool use, streaming integration.

  • API integration
  • AI applications
  • Model optimization
By AnthropicSource

Safety & validation

Karpathy Guidelines

CLAUDE.md guidelines derived from Andrej Karpathy's observations on LLM coding pitfalls.

  • Reduce overcomplication
  • Prevent dead code
  • Clearer goal scoping
By multica-aiSource

Development

Superpowers

Composable dev workflow: brainstorm, plan, subagents, and test-driven development in one framework.

  • End-to-end workflow
  • Subagent orchestration
  • Test-driven development
By obraSource

Development

Code Simplifier

Cleans up recently changed code for readability without altering behavior.

  • Readability cleanup
  • Convention alignment
  • Post-edit polish
By AnthropicSource

Design

Vercel Web Interface Guidelines

Audits UI code against 100+ accessibility and UX rules from Vercel.

  • Accessibility audit
  • UX review
  • Interface QA
By Vercel LabsSource

Testing & QA

Webapp Testing

Drive a real browser via Playwright to test local web apps end-to-end.

  • End-to-end testing
  • UI verification
  • Regression checks
By AnthropicSource

Infrastructure

Firecrawl

Web scraping, search, and browser automation for agents at scale.

  • Web scraping
  • Site crawling
  • Real-time research
By FirecrawlSource

Productivity

Caveman

Strips narration and filler to cut Claude Code output tokens by ~65%.

  • Token reduction
  • Cost control
  • Faster iterations
By Julius BrusseeSource

Security

Trail of Bits Security Skills

Professional static analysis and code auditing -- CodeQL, Semgrep, and more.

  • Static analysis
  • Code auditing
  • Vulnerability review
By Trail of BitsSource

Development

MCP Builder

Guided workflow for building high-quality MCP servers.

  • MCP server creation
  • Tool integration
  • API wrapping
By AnthropicSource

Documents

Office Document Skills

Create, edit, and analyze PDF, Word, Excel, and PowerPoint files programmatically.

  • PDF generation
  • Spreadsheet automation
  • Report building
By AnthropicSource

3D & creative

Blender Skills

Expert Claude Code skills for Blender 5.x -- geometry nodes, shaders, rigging, physics, and bpy scripting.

  • Procedural geometry nodes
  • Shader & material recipes
  • Python (bpy) automation
By ra100Source

MCP servers

Wire in your stack.

Skills teach Claude how to work; MCP servers give it live access to your tools. These are the ones we reach for first — copy the command, paste it into Claude Code.

MCP Server

GitHub MCP

Let Claude read issues, review PRs, and search across your repos without leaving the terminal.

  • Repos & Search
  • Issues & PRs
  • Actions
Setup command
claude mcp add --transport http github https://api.githubcopilot.com/mcp

MCP Server

Context7 MCP

Version-aware library docs and API signatures, injected on demand -- fewer wrong-API errors.

  • Library Docs
  • API Signatures
  • Version-aware
Setup command
claude mcp add --transport http context7 https://mcp.context7.com/mcp

MCP Server

Playwright MCP

A real browser Claude can drive to verify its own UI changes. Microsoft's official server.

  • Browser Automation
  • DOM Snapshots
  • E2E Testing
Setup command
claude mcp add playwright -- npx @playwright/mcp@latest

MCP Server

Sentry MCP

Pull production errors, stack traces, and releases straight into Claude's context.

  • Error Tracking
  • Stack Traces
  • Releases
Setup command
claude mcp add --transport http sentry https://mcp.sentry.dev/mcp

MCP Server

Supabase MCP

Database, auth, and real-time integration with Supabase -- the stack this project runs on.

  • PostgreSQL
  • Authentication
  • Real-time Subscriptions
Setup command
claude mcp add --transport http supabase "https://mcp.supabase.com/mcp?project_ref=YOUR_PROJECT_REF"

MCP Server

Blender MCP

Drive Blender from natural language -- Claude builds, edits, and renders 3D scenes in your viewport.

  • Scene & Object Control
  • Materials & Rendering
  • PolyHaven / Sketchfab Assets
Setup command
claude mcp add blender -- uvx blender-mcp

Frequently asked

The questions you'll ask anyway.

What is a Claude Code skill?
A skill is a Markdown file (SKILL.md) that lives in ~/.claude/skills/<name>/ and gives Claude Code a slash command (/name) plus instructions for what to do when invoked. Skills are local files — no plugins, no registries, no network calls. See Anthropic's skills docs.
Is this safe to install? What does it touch?
Only ~/.claude/skills/. The install command writes two directories there: handoff and build-kb. Each contains a single SKILL.md file. Nothing else on disk, no daemons, no telemetry, no network calls outside of the git clone from GitHub. The skills themselves only run the commands you see in the SKILL.md file. The auto-wrap hooks are a separate, opt-in step — nothing is registered unless you wire them into your own settings.
Do I need a paid Claude plan?
No. Skills work on any Claude Code installation, including free-tier accounts. They are local files Claude Code reads — no separate subscription, no API charges beyond your normal Claude Code usage.
What if I already have skills with the same names?
Back up first: cp -r ~/.claude/skills ~/.claude/skills.bak before running the install command. The install will overwrite same-named directories without prompting. Restore from the backup if you change your mind.
How do I update?
From inside the cloned repo: cd handoff && git pull && cp -r skills/* ~/.claude/skills/. Updates are additive and idempotent — re-running the install command also works.

Still curious? Read the full README on GitHub or create a free account to access ChatKit.