| name | readme-generator |
| description | Auto-generate concise, visually stunning README.md with project auto-scanning, badges, and 3 style modes. |
README Generator v2
Purpose
Generate concise, scannable README files that look premium on GitHub. Auto-detects project context and generates badges, features, and install instructions.
Design Philosophy
- Scan, don't read — Tables, badges, one-liners over paragraphs
- Visual-first — Center-aligned hero, shields.io badges, emoji headings
- No walls of text — Max 2-3 sentences per section, use tables/bullets
- Smart detection — Auto-reads
package.json, setup.py, VERSION, .agent/ to fill real data
Usage
Auto-generate (recommended)
python .agent/skills/readme-generator/scripts/readme_gen.py --path .
python .agent/skills/readme-generator/scripts/readme_gen.py --path . --style minimal
python .agent/skills/readme-generator/scripts/readme_gen.py --path . --style standard
python .agent/skills/readme-generator/scripts/readme_gen.py --path . --style detailed
Custom fields
python .agent/skills/readme-generator/scripts/readme_gen.py \
--path . \
--name "My Project" \
--slogan "One line that sells" \
--author "username" \
--image "https://example.com/banner.png"
Style Modes
| Mode | Lines | Best For |
|---|
minimal | ~40-60 | Libraries, packages, simple tools |
standard | ~80-120 | Most projects (default) |
detailed | ~120-180 | Complex frameworks, multi-component projects |
What It Auto-Detects
| Source | Extracts |
|---|
package.json | Name, version, license, scripts, dependencies |
setup.py / pyproject.toml | Name, version, license, dependencies |
VERSION | Version number |
LICENSE | License type |
.agent/skills/ | Skill count |
.agent/workflows/ | Workflow/agent count |
Dockerfile | Docker support |
tsconfig.json | TypeScript detection |
Key Rules for Agent
- Never write paragraphs — Use tables, bullets, or one-liners
- Hero must be centered —
<div align="center"> with title + badges + image
- Features in tables —
| Emoji Feature | One-line description |
- Install in ≤ 5 lines — Clone → install → run, no prose
- Only relevant sections — Skip Docker section if no Dockerfile
