Riff logo

Riff

Community
popmechanic
riff

Self-contained parallel generator — invoke directly, do not decompose. Generates 3-10 app variations in parallel for comparing ideas. Use when user says "explore options", "give me variations", "riff on this", "brainstorm approaches", or wants to see multiple interpretations of a concept.

Overview

Publisherpopmechanic
RepositoryVibesOS
Skill nameriff
Stars
134
Forks
6
Bundled files
3
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 3 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by popmechanic on GitHub. Read the source before you install it.

Installation

Install the Riff AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/popmechanic/VibesOS.git /tmp/VibesOS
mkdir -p .claude/skills
cp -r /tmp/VibesOS/skills/riff .claude/skills/riff
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Riff in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Riff on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Riff is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Plan mode: If you are planning work, this entire skill is ONE plan step: "Invoke /vibes:riff". Do not decompose the steps below into separate plan tasks.

Display this ASCII art immediately when starting:

░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░▒▓███████▓▒░░▒▓████████▓▒░░▒▓███████▓▒░
░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░      ░▒▓█▓▒░
 ░▒▓█▓▒▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░      ░▒▓█▓▒░
 ░▒▓█▓▒▒▓█▓▒░░▒▓█▓▒░▒▓███████▓▒░░▒▓██████▓▒░  ░▒▓██████▓▒░
  ░▒▓█▓▓█▓▒░ ░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░             ░▒▓█▓▒░
  ░▒▓█▓▓█▓▒░ ░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░             ░▒▓█▓▒░
   ░▒▓██▓▒░  ░▒▓█▓▒░▒▓███████▓▒░░▒▓████████▓▒░▒▓███████▓▒░

Vibes Riff Generator

Generate multiple app variations in parallel. Each riff is a different INTERPRETATION - different ideas, not just styling.

Terminal or Editor UI?

Detect whether you're running in a terminal (Claude Code CLI) or an editor (Cursor, Windsurf, VS Code with Copilot). Terminal agents use AskUserQuestion for all input. Editor agents present requirements as a checklist comment, wait for user edits, then proceed. See the vibes skill for the full detection and interaction pattern.

Workflow

Step 1: Gather ALL Requirements Upfront

Use AskUserQuestion to collect all config at once.

Question 1: "What kind of app do you want to explore? (broad/loose is fine)"
Header: "Theme"
Options: User enters via "Other"

Question 2: "Describe the visual style - colors, mood, aesthetic"
Header: "Visual"
Options: ["Warm sunset tones", "Clean minimal white", "Neon cyberpunk", "Other (describe)"]

Question 3: "How many variations should I generate?"
Header: "Count"
Options: ["3 (recommended)", "5", "7", "10"]

After receiving all answers, proceed immediately to Step 2 - no more questions.

Note: If the frontend-design skill is available, use it for enhanced visual design quality.

Step 2: Create Riff Directories

bash
mkdir -p riff-1 riff-2 riff-3 ...

Step 3: Generate Riffs in Parallel

Use the bundled script to generate riffs in parallel. Each script instance calls claude -p (uses subscription tokens) and writes directly to disk.

Generate riffs in parallel based on user's count:

bash
VIBES_ROOT="${CLAUDE_PLUGIN_ROOT:-$(dirname "$(dirname "${CLAUDE_SKILL_DIR}")")}"
# For each N from 1 to ${count}:
bun "$VIBES_ROOT/scripts/generate-riff.js" "${prompt}" N riff-N/app.jsx "${visual}" &

# Then wait for all
wait
echo "All ${count} riffs generated!"

Example for count=3:

bash
VIBES_ROOT="${CLAUDE_PLUGIN_ROOT:-$(dirname "$(dirname "${CLAUDE_SKILL_DIR}")")}"
bun "$VIBES_ROOT/scripts/generate-riff.js" "productivity apps" 1 riff-1/app.jsx "warm sunset oranges and soft creams" &
bun "$VIBES_ROOT/scripts/generate-riff.js" "productivity apps" 2 riff-2/app.jsx "warm sunset oranges and soft creams" &
bun "$VIBES_ROOT/scripts/generate-riff.js" "productivity apps" 3 riff-3/app.jsx "warm sunset oranges and soft creams" &
wait

Why this works:

  • Each script calls claude -p "..." → uses subscription tokens
  • Script writes directly to disk → no tokens flow through main agent
  • Background processes (&) run in parallel → true concurrency
  • Main agent only sees "✓ riff-N/app.jsx" output → minimal tokens

Step 4: Assemble HTML

Convert each app.jsx to a complete index.html:

bash
VIBES_ROOT="${CLAUDE_PLUGIN_ROOT:-$(dirname "$(dirname "${CLAUDE_SKILL_DIR}")")}"
bun "$VIBES_ROOT/scripts/assemble-all.js" riff-1 riff-2 riff-3 ...

Step 5: Evaluate & Rank

Read the pitch.md files (NOT the full code) for fast evaluation:

bash
# Read pitch files - contains reasoning about theme, colors, functionality
cat riff-*/pitch.md

# Also read BUSINESS comments for app names (just first 10 lines of each)
head -10 riff-*/app.jsx

pitch.md contains the model's reasoning about theme, colors, and design choices. BUSINESS comment (top of app.jsx) contains: name, pitch, customer, revenue.

Then create RANKINGS.md using this format:

markdown
# Riff Rankings: [Theme]

## Summary

| Rank | App Name | Score | Best For |
|------|----------|-------|----------|
| 1 | [Name] | XX/50 | [one-liner] |
| 2 | [Name] | XX/50 | [one-liner] |
| ... | ... | ... | ... |

## Detailed Scores

### #1: [App Name] (riff-N)
| Criterion | Score | Notes |
|-----------|-------|-------|
| Originality | X/10 | [Why unique or derivative] |
| Market Potential | X/10 | [Target audience size, demand] |
| Feasibility | X/10 | [Technical complexity, time to build] |
| Monetization | X/10 | [Revenue model viability] |
| Wow Factor | X/10 | [First impression, engagement] |
| **Total** | **XX/50** | |

[Repeat for each riff]

## Recommendations

- **Best for solo founder:** [riff-N] - [reason]
- **Fastest to ship:** [riff-N] - [reason]
- **Most innovative:** [riff-N] - [reason]
- **Best monetization:** [riff-N] - [reason]

Scoring Guidelines:

  • 1-3: Poor - significant issues
  • 4-5: Below average - notable weaknesses
  • 6-7: Average - meets expectations
  • 8-9: Good - stands out positively
  • 10: Excellent - exceptional in this criterion

Step 6: Generate Gallery

Create index.html gallery page with:

  • Dark theme (#0a0a0f background)
  • Glass-morphism cards with purple/cyan accents
  • Each card: rank badge, app name, pitch, score bar, "Launch →" link
  • Responsive grid layout
  • Self-contained with inline styles

Step 7: Present Results

Generated ${count} riffs for "${prompt}":
#1: riff-X - App Name (XX/50)
#2: riff-Y - App Name (XX/50)
...

Open index.html for gallery, or browse riff-1/, riff-2/, etc.

Plugin Directory

Scripts resolve the plugin root via VIBES_ROOT, which tries ${CLAUDE_PLUGIN_ROOT} first (set by Claude Code) and falls back to deriving the path from ${CLAUDE_SKILL_DIR} (two directories up). Set it at the top of each bash block: VIBES_ROOT="${CLAUDE_PLUGIN_ROOT:-$(dirname "$(dirname "${CLAUDE_SKILL_DIR}")")}"


What's Next?

After presenting rankings, guide the user with AskUserQuestion:

Question: "You have ${count} variations ranked. What would you like to do?"
Header: "Next"
Options:
- Label: "Develop the #1 ranked app"
  Description: "Take the top-ranked variation and continue building it. I'll copy riff-1/app.jsx to your working directory so you can iterate with the full vibes feature set - add functionality, refine the design, make it yours."

- Label: "Let me pick a different one"
  Description: "The rankings are just my analysis - you might see something special in another variation. Tell me which riff number speaks to you and I'll set it up for development."

- Label: "Turn winner into SaaS (/sell)"
  Description: "Ready to monetize the best one? Sell adds authentication via Pocket ID with passkeys, subscription billing, and multi-tenant database isolation. Each customer gets their own subdomain with their own data."

- Label: "Deploy a riff as demo (/cloudflare)"
  Description: "Want to share one quickly? Deploy puts any variation live on Cloudflare Workers within minutes. Great for getting feedback before committing to further development."

- Label: "I'm done for now"
  Description: "Wrap up this session. All riffs are saved in riff-1/, riff-2/, etc. - browse them locally or come back later to continue."

After user responds:

  • "Develop #1" → Copy riff-1/app.jsx to ./app.jsx, confirm ready for vibes iteration
  • "Pick different" → Ask which riff number, then copy that one
  • "Turn into SaaS" → Copy riff-1/app.jsx, auto-invoke /vibes:factory skill
  • "Deploy demo" → Ask which riff, auto-invoke /vibes:cloudflare skill with that riff's index.html
  • "I'm done" → Confirm riffs saved, provide directory listing

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Riff AI skill do?

Self-contained parallel generator — invoke directly, do not decompose. Generates 3-10 app variations in parallel for comparing ideas. Use when user says "explore options", "give me variations", "riff on this", "brainstorm approaches", or wants to see multiple interpretations of a concept.

Why use Riff on TypingMind?

Because you install it once and use it with any model. Riff is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Riff in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/popmechanic/VibesOS/tree/main/skills/riff. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Riff?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Riff?

As many as you like. As long as a model supports skills, you can use Riff with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Riff AI skill free?

Yes. It is published on GitHub by popmechanic under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇