rmb is a CLI for building and rating great READMEs. It scores any README against 50+ research-backed criteria — locally, no AI, no API calls — and bootstraps the rmb skill into your coding agents (Pi and Claude Code) so they can author README files (logo, header, and all 14 sections) that actually score well.
- Features
- Quick Start
- Why rmb?
- README Scoring
- Agent Setup
- Usage
- Documentation
- Installation
- Contributing
- License
- Acknowledgements
Most open source READMEs are mediocre. Research shows that a great README can 5× your star conversion rate — but most projects never invest the time. rmb attacks that from both ends:
- Measure what you have — There's no objective way to know if your README is good.
rmb scoregives a deterministic, research-backed score with specific, ranked recommendations — no LLM, no API calls, results in milliseconds. - Build a better one —
rmb installbootstraps the rmb skill into your coding agent, so it can author READMEs (logo, header, all 14 sections) against the same research the scorer checks.
- 📊 README scoring — Scores any README against 50+ criteria across 5 categories — no LLM required
- 🎯 Actionable feedback — Every failed check comes with a specific, ranked recommendation
- 🤖 Agent bootstrap — One command drops the rmb README skill into Pi and Claude Code
- 🔒 Fully local — Scoring is deterministic and offline; no AI, no API keys, no telemetry
- ⚙️ CI-friendly —
--jsonoutput and--check --thresholdfor failing builds on low scores
# Install the CLI
cargo install banger-readme
# Score a README
rmb score .
# Bootstrap the rmb skill into your agents (Pi + Claude Code)
rmb installrmb score evaluates READMEs against 50+ deterministic criteria — no AI, no API calls, instant results.
╔══════════════════════════════════════════════════╗
║ README Score: 85/100 [A] ║
║ "Excellent — well above average" ║
║ Top 5% of open source READMEs ║
╚══════════════════════════════════════════════════╝
┌─ Content Completeness ────── 32/35 ─────┐
│ ✅ Logo/Banner ✅ Badges ✅ One-liner │
│ ✅ Demo ✅ Features ✅ Quick Start│
│ ✅ TOC ✅ Why ✅ Install │
│ ✅ Usage ✅ API Ref ✅ Contrib │
│ ✅ License ✅ Acknowledgements │
└──────────────────────────────────────────┘
Five scoring categories:
| Category | Weight | What It Checks |
|---|---|---|
| Content Completeness | 35 pts | 14 essential sections from logo to acknowledgements |
| Visual Design | 25 pts | Logo, badges, emojis, images, collapsible sections, dark/light mode |
| Project Hygiene | 20 pts | LICENSE, CONTRIBUTING.md, CODE_OF_CONDUCT.md, SECURITY.md, CI, tests |
| Cognitive Funneling | 15 pts | Section ordering: broad→specific, one-liner before install, license at bottom |
| Anti-Patterns | −15 pts | Deductions for placeholders, duplicate titles, badge abuse, wall of text |
rmb install bootstraps the rmb README skill into your coding agents. Once installed, your agent can author READMEs — generating logos and headers, structuring all 14 sections, and following the same research rmb score checks against.
# Bootstrap into both Pi and Claude Code (default)
rmb install
# Only Pi
rmb install --pi-only
# Only Claude Code
rmb install --claude-onlyIt fetches the bundled banger-readme plugin from GitHub and installs it to:
| Target | Location |
|---|---|
| Pi coding harness | ~/.pi/agent/skills/banger-readme/ |
| Claude Code | ~/.claude/plugins/cache/ (skills symlinked into ~/.claude/skills/) |
Re-running rmb install replaces the existing installation, so it doubles as an update command.
🔍 Scoring a README
# Score the README in the current directory
rmb score .
# Score a remote README
rmb score https://raw.githubusercontent.com/user/repo/main/README.md --no-hygiene
# Machine-readable JSON output
rmb score . --json
# CI mode: fail if score below threshold
rmb score . --check --threshold 70🤖 Bootstrapping into agents
# Install the rmb skill into both Pi and Claude Code
rmb install
# Only one target
rmb install --pi-only
rmb install --claude-only| Document | Description |
|---|---|
| Scoring Engine Design | Full architecture and detection rules for rmb score |
| README Best Practices Research | Index of all research on what makes a great README |
| Why READMEs Matter | The data, psychology, and philosophy |
| Anatomy of a README | The 14-section structure in detail |
| Project Hygiene | License, contributing, CoC, security, tests |
| Aesthetics & Visual Design | Formatting, visual elements, linkification |
| Emerging Patterns | Trends for 2024–2026 |
| Anti-Patterns | What repels users and common mistakes |
| Exemplary READMEs | Curated examples by category |
| Examples Audit | How 31 production READMEs rank against criteria |
git clone https://github.com/kerryhatcher/banger-readme
cd banger-readme
cargo build --releaseThe binary will be at target/release/rmb.
cargo install banger-readmecargo test
cargo clippy -- -D warnings
cargo fmt -- --check- Questions? Open a GitHub Discussion
- Bug? File an issue
- Security issue? See SECURITY.md
We welcome contributions! See CONTRIBUTING.md for development setup, project structure, code standards, and the pull request process.
MIT — see LICENSE for details.
Built with research from:
- Awesome README — 21.2k stars, curated list of 100+ exemplary READMEs
- Art of README — Foundational essay on README psychology
- Readme Driven Development — Tom Preston-Werner
- Daytona's 4,000-star launch guide
- Elegant READMEs — Yegor Bugayenko