EXPAND vague prompts into precise, platform-optimized instructions. Detects target platform (Claude, GPT, Gemini, Midjourney, Sora, etc.) and applies appropriate prompting patterns. Use when user says "coggle", "expand", "improve prompt", "make better".
/plugin marketplace add lisaross/coggle/plugin install lisaross-coggle@lisaross/coggleThis skill inherits all available tools. When active, it can use any tool Claude has access to.
templates/claude.mdtemplates/codex.mdtemplates/copilot.mdtemplates/flux.mdtemplates/gemini.mdtemplates/grok.mdtemplates/image-gen.mdtemplates/perplexity.mdtemplates/video-gen.mdTransform vague prompts into precise, platform-optimized instructions.
Users invoke this skill by saying:
Detect target platform from context clues:
| Clue | Platform |
|---|---|
| "for Claude", "Claude Code", "system prompt" | Claude |
| "GPT", "ChatGPT", "OpenAI", "Codex" | OpenAI |
| "Gemini", "Google AI" | Gemini |
| "Perplexity", "search", "research", "citations" | Perplexity |
| "Copilot", "GitHub", "code completion" | GitHub Copilot |
| "Grok", "xAI", "X AI" | Grok |
| "Midjourney", "MJ", "--ar", "DALL-E", "image" | Image Generation |
| "Flux", "FLUX.1", "Black Forest" | Flux |
| "Sora", "Runway", "video", "animation" | Video Generation |
| "Nano Banana", "Higgsfield" | Nano Banana Pro |
If unclear, ask user: "What platform is this prompt for?"
Identify target from prompt context or ask user.
Adapt elements based on platform:
| Element | Text AI | Image AI | Video AI |
|---|---|---|---|
| Persona | Role definition | Style/artist reference | Director's vision |
| Requirements | Deliverables | Visual elements | Shots/scenes |
| Examples | Reference outputs | Reference images | Reference clips |
| Context | Background info | Scene setting | Narrative context |
| Instructions | Step-by-step | Composition notes | Storyboard |
| Specifications | Output format | Parameters (--ar, --v) | Duration, resolution |
| Evaluation | Success criteria | Visual quality checks | Motion coherence |
Define the AI's role or style:
Specify deliverables:
Provide concrete references:
Add relevant background:
Break down the task:
Define format and structure:
Specify success criteria:
Reference platform-specific patterns from ./templates/[platform].md:
| Platform | Best For | Key Pattern | Template File |
|---|---|---|---|
| Claude | Complex reasoning, coding | Role + Context + Task + Constraints | claude.md |
| Codex/GPT | Code generation, chat | Delimiters + Examples | codex.md |
| Gemini | Multimodal, reasoning | Few-shot examples | gemini.md |
| Perplexity | Research, citations | Search scope + Recency + Sources | perplexity.md |
| Copilot | Code completion | General goal + Specific requirements | copilot.md |
| Grok | Conversational, real-time | Role + Task + Format | grok.md |
| Image Gen | Midjourney, DALL-E | Descriptive phrases + Parameters | image-gen.md |
| Flux | Photorealistic images | Natural language descriptions | flux.md |
| Video Gen | Sora, Runway, Kling | Storyboard + Camera direction | video-gen.md |
When expanding a prompt, detect the target platform and apply the appropriate template structure.
After expansion, offer:
.prompts/[name].md## Coggle!
**Original:**
> [input]
**Platform:** [detected platform]
**Expanded:**
[Full expanded prompt in platform-appropriate format]
---
**Templates:** See `./templates/` for platform-specific patterns.
**Next Steps:**
- **Run** - Execute this prompt
- **Save** - Save to `.prompts/[name].md`
- **Refine** - Make adjustments
| Anti-Pattern | Example | Fix |
|---|---|---|
| Vague verbs | "help", "something", "stuff" | Specific actions |
| Missing audience | No target reader defined | Add audience context |
| Implicit assumptions | Unstated requirements | Make explicit |
| Wall of text | No structure | Add sections/steps |
| Conflicting instructions | Contradictory asks | Resolve conflicts |
| Missing format | No output spec | Define structure |
This skill is successful when:
Text AI (Claude):
Before: "Write something about marketing"
After: "Act as a marketing strategist with 10+ years B2B SaaS experience.
Write a 500-word blog post about email marketing best practices for
startup founders. Include 3 actionable tips with examples. Format as:
Hook -> Problem -> Solution -> Examples -> CTA.
Tone: Professional but approachable."
Image AI (Midjourney):
Before: "sunset over mountains"
After: "Golden hour sunset over snow-capped mountain peaks, dramatic
god rays through clouds, alpine lake reflection in foreground,
cinematic composition, in the style of Ansel Adams, photorealistic,
--ar 16:9 --v 6 --q 2"
Video AI (Sora):
Before: "person walking in city"
After: "Cinematic tracking shot following a silhouetted figure walking
through rain-soaked Tokyo streets at night, neon reflections on wet
pavement, bokeh lights in background, film noir aesthetic, slow motion,
camera dollies forward maintaining subject in center frame,
4K resolution, 24fps, 10 seconds duration"
Research AI (Perplexity):
Before: "tell me about quantum computing"
After: "What are the latest breakthroughs in quantum error correction?
Focus on peer-reviewed research from 2024-2025.
Summarize the top 3 developments with inline citations.
Include practical implications for each breakthrough."
Code AI (GitHub Copilot):
Before: "make a retry function"
After: "Write a TypeScript function that retries async operations with
exponential backoff.
Requirements:
- Accept a function, max retries (default 3), initial delay (default 1000ms)
- Double delay after each retry
- Return successful result or throw after max retries
- Log each attempt with attempt number"
Conversational AI (Grok):
Before: "analyze this startup"
After: "Role: Skeptical tech analyst who's seen too many hype cycles
Task: Evaluate the claims in this startup pitch
Format: Bullet points with a reality-check score (1-10) for each claim
Be direct, flag red flags, and rate overall viability."
Prompt Expander skill | Last Updated: 2025-12-26
Creating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems. Create original algorithmic art rather than copying existing artists' work to avoid copyright violations.
Applies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.
Create beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations.