bradygaster Context
Use this skill when reviewing README files, getting-started guides, or any documentation that serves as a developer's first experience with a project. These patterns reduce friction at key decision points: "Can I use this?", "Should I use this?", "How do I use this?", "Will this work for me?"
Patterns
1. Surface Project Status Early
Place experimental/beta/production status within the first 3 lines of the README, ideally right after the tagline. Use visual markers (⚠️, 🚧, ✅) to signal maturity level.
Why: Developers scanning documentation need to self-select immediately. Burying "this is experimental" disclaimers at the bottom wastes their time and creates frustration.
Example Placement:
# ProjectName
> Tagline describing what this is
> ⚠️ **Active Experiment**: This is a learning journey, not production software.
First paragraph explaining the project...
2. Prerequisites Before Quick Start
List required tools, subscriptions, accounts, and SDKs before any clone/fork/install instructions. Include a verification step so developers can confirm their environment is ready.
Components:
- Tools (IDE, CLI, runtime)
- Accounts/subscriptions (cloud services, API keys)
- Minimum versions
- Verification command or test
Why: Prevents the "I followed all the steps but nothing works" scenario. Developers know upfront what's required and can test their setup before investing time.
3. Platform-Appropriate Command Examples
Match your command syntax to your primary developer audience's expected environment. For .NET/C# projects, show Windows PowerShell first. For Node.js/Python projects, show Unix shell first.
Pattern:
# Primary audience command
Remove-Item -Recurse -Force .git # Windows PowerShell
# Alternative: rm -rf .git for Linux/macOS
Why: Reduces cognitive load. Developers work in their native environment and shouldn't have to mentally translate commands.
4. Lead with Roles, Not Counts
When describing systems with multiple components/agents/services, emphasize roles and capabilities over implementation counts. "Specialists for testing, deployment, and security" is clearer than "23 specialized components."
Why: Implementation details change. The conceptual model should be stable. Leading with counts makes the framework feel arbitrary.
Before: "21 specialized AI agents"
After: "Specialized AI agents covering testing, security, deployment, and databases"
5. Bridge the Setup-to-Usage Gap
Add a "What Happens Next?" section after getting-started instructions. Show what the developer will see, what gets created, and how to verify the first successful interaction.
Structure:
- What to do first (first command/interaction)
- What output to expect
- How to know it worked
- Where to go next
Why: The gap between "I installed this" and "I'm productively using this" is where most abandonment happens. Explicit guidance bridges that gap.
Examples
Good Status Visibility
# ProjectName
> An AI-powered code generator for REST APIs
> 🚧 **Beta Software**: Core features work, but expect breaking changes and rough edges.
ProjectName helps you...
Prerequisites Section
## Prerequisites
Before getting started, ensure you have:
- Node.js 18+ installed
- GitHub account with Copilot subscription
- Azure subscription (free tier works)
**Verify your setup:**
```bash
node --version # Should show v18.0.0 or higher
gh auth status # Should show "Logged in to github.com"
### What Happens Next Section
```markdown
## What Happens Next?
After cloning the repo:
1. Run `npm install` to install dependencies
2. Run `npm start` to launch the dev server
3. Open http://localhost:3000 — you should see the welcome screen
4. Try creating your first API: `npm run generate -- --name todos`
You'll see output like:
✓ Generated models/todo.ts ✓ Generated routes/todos.ts ✓ Updated index.ts
The API is now available at GET/POST http://localhost:3000/api/todos
Anti-Patterns
- Burying disclaimers — "This is experimental" hidden in a footer or license section
- Assuming prerequisites — Not listing required tools because "everyone has Node installed"
- Universal commands — Showing only
rm -rfon a Windows-focused project - Setup without verification — No way to test if the setup actually worked
- Implementation-first framing — Leading with "327 modules" instead of "what it does"
- Skipping the gap — Going from installation directly to API reference without showing first usage