Skip to main content

Reality-First Development Protocol - Prevents AI hallucination and ensures spec-driven development

Project description

RFD Protocol - Reality-First Development

Stop AI hallucination. Ship working code.

CI Pipeline Release Pipeline PyPI version Python 3.8+ License: MIT

The Problem We Solve

48% of AI-generated code contains hallucinations - false claims, non-existent functions, or broken implementations. Developers waste countless hours debugging phantom code, losing context between sessions, and watching projects drift into chaos.

RFD enforces reality at every step. No more "I implemented that feature" when nothing works. No more mock data pretending to be production code. No more losing track of what you were building.

What is RFD?

RFD (Reality-First Development) is a development protocol that makes AI hallucination physically impossible through continuous reality validation. Not just managing the LLM but also the human to help prevent squirrel brain and SDLC drift by enforcing concrete reality checkpoints. Instead of trusting AI claims about what was implemented, RFD validates the actual code runs, tests pass, and features work. It's not just another tool - it's a fundamental shift in how we build software with AI.

Core Guarantees

Zero Hallucination - Every claim is validated against running code
Persistent Context - Never lose your place, even across restarts
Enforced Focus - Can't drift from specified features
Real Code Only - No mocks, stubs, or placeholder implementations
Universal Compatibility - Works with any language, any framework

📚 Documentation

Quick Start (90 Seconds)

# Install RFD
pip install rfd-protocol

# Initialize your project  
cd your-project
rfd init --wizard

# Start building
rfd session start my_feature
rfd build
rfd validate
rfd checkpoint "Feature complete"

That's it. RFD now guards your development.

How RFD Works

1. Specification Lock

# PROJECT.md defines what can be built
features:
  - id: user_auth
    acceptance: "Users can register and login"
    status: pending

2. Reality Enforcement

# AI claims: "I implemented user authentication"
$ rfd validate

❌ Reality Check Failed:
  - No /register endpoint found
  - No /login endpoint found  
  - 0 of 5 tests passing
  - Database table 'users' does not exist

3. Progress Tracking

$ rfd status

📊 Project Status
━━━━━━━━━━━━━━━━━━━━
Features: 1 pending, 0 complete
Current: user_auth (0% complete)
Next: Create user registration endpoint

Last Valid Checkpoint: 2 hours ago
"Database schema created"

Complete Workflow Commands

Initialization & Setup

rfd init                    # Basic project setup
rfd init --wizard          # Interactive setup (recommended)
rfd init --from-prd doc.md # Initialize from requirements doc
rfd init --mode brownfield # For existing projects

Spec-Kit Style Commands (NEW v3.0!)

rfd speckit constitution   # Create immutable project principles
rfd speckit specify       # Define detailed specifications
rfd speckit clarify      # Identify ambiguities
rfd speckit plan         # Create implementation plan
rfd speckit tasks        # Generate task breakdown

Development Workflow

rfd session start <feature>  # Begin feature work
rfd build                    # Run build process
rfd validate                 # Validate implementation
rfd checkpoint "message"     # Save progress
rfd session end             # Complete feature

Analysis & Review

rfd check                   # Quick status check
rfd status                  # Detailed project status
rfd analyze                # Cross-artifact consistency check (NEW!)
rfd dashboard              # Visual progress dashboard
rfd spec review            # Review current specification

State Management

rfd revert                  # Revert to last checkpoint
rfd memory show            # Display context memory
rfd memory reset           # Clear context (careful!)

Advanced Features (NEW in v3.0)

SQLite with WAL Mode

  • Write-Ahead Logging for better concurrency
  • Persistent memory across all sessions
  • Automatic crash recovery
  • Zero configuration required

Cross-Artifact Analysis

$ rfd analyze

CROSS-ARTIFACT ANALYSIS REPORT
════════════════════════════════
📋 SPEC ALIGNMENT
   All features aligned with specification

📝 TASK CONSISTENCY  
   All tasks consistent with feature status

🔌 API IMPLEMENTATION
  Coverage: 100.0%
   All endpoints implemented

🧪 TEST COVERAGE
  Coverage: 87.5%
   All acceptance criteria covered

Integration with AI Tools

Claude Code Configuration

RFD automatically configures Claude Code to prevent hallucination:

# Tell Claude to continue your project
"Continue the RFD session"

# Claude automatically:
$ rfd check
> Current feature: user_auth
> Last checkpoint: "Created User model"  
> Next task: Implement registration endpoint

# Claude cannot fake progress:
$ rfd checkpoint "Added authentication" Cannot checkpoint - validation failing

Custom AI Integration

For other AI tools, enforce this workflow:

  1. Read PROJECT.md for specifications
  2. Check .rfd/context/current.md for current task
  3. Run rfd validate after every change
  4. Only checkpoint when validation passes

Project Configuration

PROJECT.md Schema

RFD uses a flexible, extensible schema:

---
# Required Fields
name: "Your Project"
description: "What it does"
version: "1.0.0"

# Technology Stack (extensible)
stack:
  language: python
  framework: fastapi
  database: postgresql
  # Add any custom fields:
  runtime: python-3.11
  deployment: kubernetes
  monitoring: prometheus

# Validation Rules
rules:
  max_files: 50
  max_loc_per_file: 500
  must_pass_tests: true
  no_mocks_in_prod: true
  min_test_coverage: 80

# Features
features:
  - id: core_api
    description: "RESTful API"
    acceptance: "All endpoints return correct data"
    status: pending
    priority: high
    depends_on: []

# Constraints
constraints:
  - "Response time < 200ms"
  - "Support 10k concurrent users"
---

Project Architecture

Our repository follows modern Python packaging standards:

rfd-protocol/
├── src/rfd/                  # 🎯 MAIN PACKAGE
│   ├── __init__.py
│   ├── cli.py               # Command-line interface
│   ├── rfd.py              # Core orchestration
│   ├── validation.py       # Hallucination detection
│   ├── session.py          # Session management
│   ├── build.py            # Build automation
│   ├── spec.py             # Specification management
│   ├── analyze.py          # Cross-artifact analysis (NEW!)
│   ├── db_utils.py         # WAL mode database (NEW!)
│   └── templates/          # Project templates
│
├── tests/                   # 🧪 TEST SUITE
│   ├── unit/
│   ├── integration/
│   └── system/
│
├── docs/                   # 📚 DOCUMENTATION
│   ├── RFD_WALKTHROUGH.md # Complete guide (NEW!)
│   └── [other docs]
│
└── .github/workflows/      # 🚀 CI/CD
    ├── ci.yml
    └── release.yml        # Auto PyPI publishing

Installation Options

Global Install (Recommended)

pip install rfd-protocol

Project-Specific Install

cd your-project
python -m venv venv
source venv/bin/activate
pip install rfd-protocol

Development Install

git clone https://github.com/kryptobaseddev/rfd-protocol.git
cd rfd-protocol
pip install -e .

Language Support

RFD works with any technology stack:

  • Python: FastAPI, Django, Flask
  • JavaScript/TypeScript: Express, Next.js, React
  • Go: Gin, Echo, Fiber
  • Rust: Actix, Rocket, Axum
  • Java/Kotlin: Spring Boot
  • C/C++: Any build system
  • Ruby: Rails, Sinatra
  • PHP: Laravel, Symfony
  • And 20+ more...

Real-World Impact

Before RFD

  • 48% hallucination rate
  • Lost context after restarts
  • Endless debugging of AI mistakes
  • Projects that never ship

After RFD

  • 0% hallucination rate
  • Perfect context persistence
  • Only real, working code
  • Consistent project delivery

Testing & Development

Running Tests

# All tests
pytest

# By category
pytest -m unit           # Fast unit tests
pytest -m integration    # Integration tests  
pytest -m system         # End-to-end tests

# With coverage
pytest --cov=src/rfd --cov-report=html

Code Quality

# Linting
ruff check src tests

# Formatting
ruff format src tests

# Type checking
mypy src --ignore-missing-imports

Troubleshooting

"Feature not in PROJECT.md"

You tried to work on an undefined feature. Edit PROJECT.md first.

"Validation failed"

rfd validate --verbose  # See detailed errors
rfd build              # Fix build issues first

"Lost context"

rfd check                      # Current status
cat .rfd/context/current.md   # Session details

Debug Mode

export RFD_DEBUG=1
rfd validate                    # Verbose output

Contributing

We welcome contributions! RFD uses itself for development:

  1. Fork the repository
  2. Run rfd init in your fork
  3. Create feature in PROJECT.md
  4. Use RFD workflow to implement
  5. Submit PR when rfd validate passes

Development Setup

git clone https://github.com/kryptobaseddev/rfd-protocol.git
cd rfd-protocol
pip install -e ".[dev]"
pytest  # Run tests

Support

Version History

  • v3.0.0: SQLite WAL mode, cross-artifact analysis, spec-kit feature parity, comprehensive walkthrough
  • v2.3.0: Mock detection, critical fixes, session persistence improvements
  • v2.0.0: Spec generation, gated workflow, AI validation
  • v1.0.0: Production release with modern Python packaging and full documentation

License

MIT License - see LICENSE


Built with RFD - This project dogfoods its own reality-first methodology.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

rfd_protocol-4.1.0.tar.gz (101.8 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

rfd_protocol-4.1.0-py3-none-any.whl (93.4 kB view details)

Uploaded Python 3

File details

Details for the file rfd_protocol-4.1.0.tar.gz.

File metadata

  • Download URL: rfd_protocol-4.1.0.tar.gz
  • Upload date:
  • Size: 101.8 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.13

File hashes

Hashes for rfd_protocol-4.1.0.tar.gz
Algorithm Hash digest
SHA256 eba2670c5bcce8652e85a937fc9aa54e5137313d42b97e5570389b227d6103fd
MD5 8926e0793598b294a1a88965d33bee8d
BLAKE2b-256 7297a5271fb2b4181c01ac793e3c1e6b0e35437e3b5fb61fb5081f77a910519c

See more details on using hashes here.

File details

Details for the file rfd_protocol-4.1.0-py3-none-any.whl.

File metadata

  • Download URL: rfd_protocol-4.1.0-py3-none-any.whl
  • Upload date:
  • Size: 93.4 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.13

File hashes

Hashes for rfd_protocol-4.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 824cb5e4c92ffeca4386287ef3315249cbb4b292a697c399e6dd2789a4ec510e
MD5 6048040be2829a495da3fc3ca555e23f
BLAKE2b-256 2c97aeaebbd469464c73c119415e5a6b421a387ae5a45eb1333398d2594f73ea

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page