Hand-crafted Claude skills for founders who do their own design, writing, ops, and everything in between.
A growing collection of Claude skills built for founders running lean — the kind who are the designer and the writer and the ops person, not someone delegating each hat to a team.
Each skill is designed to work on its own, drops cleanly into any Claude conversation, and stays brand-agnostic so you can fit it to whatever you're building.
| Skill | What it does |
|---|---|
mini-saas-landing-page |
Builds a card-grid section for any SaaS landing page — features, how-it-works, use cases, or "built for [role]" — using a proven, brand-agnostic header + 2/3/4-card layout. |
articulation-skill |
Helps founders communicate with clarity across three modes: talking with your team about craft, aligning with stakeholders, and creating public content. |
A reusable layout for the most common card-grid section on a SaaS landing page: a header with title and tagline, followed by 2–4 cards with a visual on top and short copy below, optionally followed by a CTA strip.
Use it when you're:
- Designing a features section for a new product
- Sketching a "how it works" block for an existing landing page
- Wireframing a use-cases or "built for [role]" section
- Adapting an existing layout to a new brand without redesigning from scratch
What's inside:
- Three layout variants (2-card, 3-card default, 2×2 grid) chosen by how many capabilities you have
- Five reusable visual patterns (mock product window, with/without comparison, tabbed dashboard, input→output flow, pill+connector decoration) with ASCII sketches
- A complete HTML template with CSS custom properties for brand tokens — swap your brand by editing ~10 variables
- Responsive breakpoints, dark mode, and accessibility baked in
- Copy rules for concrete-outcome titles and filler-free descriptions
Trigger phrases: "features section", "capabilities grid", "how it works", "what we do", "build me a landing page section"
A communication skill for founders who wear multiple hats. Based on the idea that founders live in three worlds — the workshop, the stakeholder room, and the public timeline — and need to move between them without losing credibility in any.
Use it when you're:
- Writing an incident postmortem, PR description, or design critique
- Drafting a board update, investor email, or co-founder conversation
- Shipping a LinkedIn post, X thread, blog post, or conference talk
- Reviewing writing that feels too long, too vague, or too generic
What's inside:
- A "three hats" model (Builder / Leader / Voice) that maps any communication task to the right framework
- Three frameworks: the Debuggable Explanation, the Pyramid Principle, and the Content Hook
- A compression pass that typically cuts 30–40% of a draft's length with no loss of meaning
- "Idea Legos" — reusable building blocks (tradeoffs, war stories, metrics, opportunity cost) for when a draft feels thin
- "Founder Lenses" — recurring mental models that help you find the point you're making, not just how to say it
- Worked examples across engineering, design, ops, and content contexts
Trigger phrases: "help me explain this", "draft a post about", "review my writing", "make this clearer", "shorten this", "prep me for this meeting"
Clone the repo and run the installer:
git clone https://github.com/YOUR-USERNAME/founder-forge.git
cd founder-forge
./install.sh
That's it. The installer symlinks each skill into ~/.claude/skills/, so Claude Code picks them up on the next session. Because it's a symlink, upgrading later is just git pull — no re-install needed.
Install options:
./install.sh # install all skills
./install.sh mini-saas-landing-page # install just one
./install.sh --list # list available skills
./install.sh --uninstall <name> # remove an installed skill
./install.sh --help # full help
If you're using claude.ai instead of Claude Code:
- Download the
.skillfile for the skill you want from the Releases page. - Open Claude → Settings → Capabilities → Skills → Upload skill.
- Select the
.skillfile.
The skill triggers automatically when you ask Claude something it matches. Note that claude.ai doesn't auto-upgrade — you'll need to re-upload the .skill file when a new release ships. Watch this repo on GitHub to get notified of new releases.
If you're new to Claude skills, read Anthropic's skills documentation first.
Claude Code users: cd into your clone and run git pull. The installer uses symlinks, so the pull alone is the upgrade.
claude.ai users: download the new .skill file from the latest release and re-upload it. Anthropic doesn't currently support auto-upgrade for manually-uploaded skills.
See CHANGELOG.md for what's changed in each release.
founder-forge/
├── README.md ← you are here
├── CHANGELOG.md ← what changed, per skill
├── LICENSE ← MIT
├── install.sh ← one-command Claude Code installer
├── mini-saas-landing-page/
│ ├── SKILL.md
│ ├── assets/
│ │ └── template.html
│ └── references/
│ ├── visual-patterns.md
│ └── layout-structures.md
└── articulation-skill/
├── SKILL.md
└── references/
├── idea-legos.md
├── founder-lenses.md
└── examples.md
Every skill follows the same shape: a SKILL.md with the main instructions, optional assets/ for files the skill outputs (templates, icons), and optional references/ for documentation the skill reads as needed.
Spotted something that could be sharper? Found an edge case where a skill breaks? Open an issue or send a PR. I especially welcome:
- Better examples — real ones from your own work, not hypotheticals
- Bug reports — cases where a skill triggered when it shouldn't have, or didn't when it should have
- New skill proposals — if you've built something for your own founder workflow and it feels reusable, I'd love to see it
Please don't send PRs that:
- Add new skills without discussing the scope in an issue first (easier to get aligned upfront)
- Bundle third-party content without a compatible license
- Contain anything brand-specific or customer-specific from your own work
Three rules I try to hold every skill in this repo to:
- Brand-agnostic. No skill should lock you into one visual style, voice, or industry. If a skill only works for one product, it doesn't belong here.
- Compressed. Every skill is the shortest useful version of itself. If a section doesn't earn its space, it gets cut.
- Honest about tradeoffs. Every skill names what it's good at and what it's not good at. Skills that pretend to work everywhere work nowhere.
If you build on top of anything here, I'd love to hear about it.
Licensed under the MIT License. Use it, modify it, ship it, sell things built with it — just keep the copyright notice.
Built with Claude. If you find these skills useful, the best thanks is to tell another founder about them.