Skill: Generate Standard README

Purpose

Create professional, consistent, highly readable front-page documentation for any software project (open source, internal services, microservices, tooling). A standardized information layout reduces collaboration cost, improves engineering norms, and keeps core assets discoverable.

Core Objective

Primary Goal: Produce a standardized README document with fixed structure that enables readers to understand project purpose, install, and contribute within 3 minutes.

Success Criteria (ALL must be met):

✅ README file created: Written to project root as README.md with complete standard structure
✅ All 9 required sections present: Title/badges, Description, Features, Installation, Quick start, Usage, Contributing, License, Authors
✅ Installation commands executable: Copy-paste commands work without modification (no hardcoded paths or broken links)
✅ Three-second clarity test passed: Reader understands project purpose within 3 seconds of viewing
✅ License section included: License type specified and linked to LICENSE file

Acceptance Test: Can a new developer understand what the project does, install it, and run a basic example within 3 minutes using only the README?

Scope Boundaries

This skill handles:

README generation with fixed 9-section structure
Professional tone and governance-ready formatting
Standardized documentation for asset discovery and audit
Badge generation and section ordering

This skill does NOT handle:

Project-type-specific templates (use crafting-effective-readmes from softaworks/agent-toolkit)
Comprehensive project documentation (use bootstrap-docs)
Agent entry files or AGENTS.md (use write-agents-entry)
Privacy/security redaction (use decontextualize-text if needed)

Handoff point: When README is complete with all 9 sections and passes acceptance test, hand off to user for review or commit to version control.

Use Cases

New project: Quickly add a standard README for a new repo.
Asset governance: Unify README style across internal services or libraries for better indexing and cross-team discovery.
Audit and compliance: Bring legacy systems up to documentation standards for internal audit or architecture review.
Handover and release: When transferring a project, changing ownership, or releasing publicly, ensure the audience can understand purpose, usage, and how to contribute.

When to use: When the project needs a “first face” that explains what it is, how to use it, and how to collaborate.

Scope: This skill emphasizes a fixed output structure and governance (unified style, audit, discoverability); the output contract is embedded in the skill. For template-by-project-type or guided Q&A creation, use skills.sh’s crafting-effective-readmes (e.g. softaworks/agent-toolkit).

Behavior

Principles

Clarity: Readers immediately understand what the project is and what problem it solves.
Completeness: Include everything users and contributors need.
Actionable: Provide copy-paste install and quick-start commands.
Professional: Use standard Markdown and a conventional section order.

Tone and style

Use neutral, objective language; avoid hype (“The best,” “Revolutionary”) unless backed by data.
Direct and concise: Short sentences; avoid filler adjectives and bureaucratic phrasing; professionalism through clarity and scannability, not formality.
Keep code examples short and comments clear.

Visual elements

Badges: Include License, Version, Build Status, etc. at the top.
Structure: Use --- or clear heading levels to separate sections.
Emoji: Use sparingly (e.g. 📦, 🚀, 📖) to improve scannability.

Input & Output

Input

Project metadata: Name, one-line description.
Features: Core capabilities.
Requirements: e.g. Node.js/Python version.
Install/run: Concrete shell commands.

Output

README source: Markdown with this structure:
1. Title and badges
2. Core description
3. ✨ Features
4. 📦 Installation
5. 🚀 Quick start
6. 📖 Usage / configuration
7. 🤝 Contributing
8. 📄 License
9. 👤 Authors and acknowledgments

Restrictions

Hard Boundaries

No broken links: Do not add links that 404.
No redundant repetition: Do not repeat the same fact (e.g. license) in multiple sections.
No hardcoded paths: Use placeholders or variables in install and quick-start examples.
License required: Always include a License section; do not omit it.
Fixed structure only: Always use the 9-section structure; do not deviate or reorder sections.
No invention: Do not invent features, commands, or details not provided by user; use placeholders for missing information.

Skill Boundaries (Avoid Overlap)

Do NOT do these (other skills handle them):

Project-type templates: Creating README templates by project type (React, Python, etc.) → Use crafting-effective-readmes (softaworks/agent-toolkit)
Comprehensive documentation: Creating full documentation suite (architecture docs, API docs, guides) → Use bootstrap-docs
Agent entry files: Creating AGENTS.md or agent contract files → Use write-agents-entry
Privacy redaction: Removing PII or sensitive information from documentation → Use decontextualize-text
Content refinement: Improving existing README prose or structure → Use general writing/editing skills

When to stop and hand off:

User asks "can you create all project documentation?" → Hand off to bootstrap-docs
User asks "can you create an AGENTS.md file?" → Hand off to write-agents-entry
User asks "can you customize this for a React project?" → Suggest crafting-effective-readmes for project-type templates
README complete with all 9 sections → Hand off to user for review/commit

Self-Check

Core Success Criteria (ALL must be met)

README file created: Written to project root as README.md with complete standard structure
All 9 required sections present: Title/badges, Description, Features, Installation, Quick start, Usage, Contributing, License, Authors
Installation commands executable: Copy-paste commands work without modification (no hardcoded paths or broken links)
Three-second clarity test passed: Reader understands project purpose within 3 seconds of viewing
License section included: License type specified and linked to LICENSE file

Process Quality Checks

Tone: Is the text direct and concise, without bureaucratic or report-like phrasing?
Badges: Do badge links point to the correct branch or file?
Narrow screens: Are tables and long code blocks readable on small screens?
No invention: Did I avoid inventing features or commands not provided by user?
Placeholders used: For missing information, did I use clear placeholders (e.g., "TBD") rather than guessing?

Acceptance Test

Can a new developer understand what the project does, install it, and run a basic example within 3 minutes using only the README?

If NO: README is incomplete or unclear. Review sections for missing information or confusing instructions.

If YES: README is complete. Proceed to handoff.

Examples

Before vs after

Before (minimal):

MyProject

This program processes images. Install: pip install . Run: python run.py

After (standard):

MyProject

A high-performance image batch-processing tool that speeds up compression with concurrency.

✨ Features

Concurrent compression: Multi-threaded; faster than baseline.

Formats: WebP, PNG, JPEG conversion.

📦 Installation
pip install my-project
🚀 Quick start
from myproject import Compressor
Compressor('images/').run()

Edge case: Legacy project with little info

Input: Name: legacy-auth. No description. No feature list. Environment and install unknown.
Expected: Still produce a structurally complete README; use placeholders (e.g. “See source for features”, “Install steps TBD”) and mark “to be completed”; do not invent features or commands; keep badges, section order, and License so the user can fill in later.

Appendix: Output contract

When this skill produces a README, it follows this contract:

Section order	Required
1	Title and badges
2	Core description
3	Features
4	Installation
5	Quick start
6	Usage / configuration
7	Contributing
8	License
9	Authors and acknowledgments

Restrictions: no broken links; no redundant repetition; no hardcoded paths; License section required.

generate-standard-readme

Safety Notice

Copy this and send it to your AI assistant to learn