DevObin

AI Engineering Context Compiler & Prompt Architect

DevObin is a CLI tool that analyzes your software project and generates production-ready engineering prompts for AI coding agents (Claude, Codex, Cursor, GPT, etc.).

DevObin does NOT write code. It builds the perfect context so the AI writes the best code.

---

What It Does

You: "Add user authentication to my Django app"

DevObin scans your project → detects Django, REST API, PostgreSQL
    ↓
DevObin generates: authentication_prompt.md
    ↓
Give the file to Claude/Codex → they implement it perfectly

Key difference: DevObin reads YOUR actual codebase and produces prompts based on YOUR real architecture — not generic templates.

---

Installation

From PyPI (Recommended)

pip install devobin

From Source

git clone https://github.com/mobinhasanghasemi/devobin.git
cd devobin
pip install -e .

Requirements

• Python 3.12+
• An AI provider API key (OpenAI, Anthropic, Google, or Ollama)

---

Quick Start

# 1. Launch DevObin
devobin

# 2. Connect to your AI provider
/connect

# 3. Describe what you need
Build a React dashboard with dark mode

# 4. Get your prompt
# → dashboard_prompt.md is created
# → Give it to Claude, Codex, or Cursor

---

Features

Smart Project Analysis

Scans your workspace, detects technologies (80+ frameworks, languages, databases), and understands your architecture.

Knowledge Graph

Builds a structured model of your project with relationships between technologies, patterns, and domains.

Dynamic Context Selection

Unlike static prompt generators, DevObin selects only relevant sections based on your intent:

• Bug fix → Security, Architecture, Tech Stack
• New feature → Full context with Design System, Components
• Performance → Optimization tips, Caching patterns

20 Prompt Sections

Each prompt includes structured sections:

Section	Description
Read Project First	Instructs AI to read code before writing
Role	Defines AI's expertise level
Tech Stack	Detected technologies
Constraints	DO / DO NOT rules
Deliverables	Expected output
Code Quality	Standards and conventions
Accessibility	WCAG AA compliance
SEO	Meta tags, Open Graph
Performance	Optimization requirements
Security	Security rules
And 10 more...	Architecture, Testing, Design System, etc.

Prompt Optimization

Optional optimization with three modes:

Mode	Description	Latency
none	No optimization	0
quick	Contradiction check + 1 mutation	~2s
full	Contradiction + mutation + critique	~10s

Plugin System

Extend DevObin without modifying core code:

from devobin.plugins import Plugin
from devobin.tools import ToolDefinition, ToolResult

class MyPlugin(Plugin):
    name = "django-expert"
    version = "1.0.0"

    def register_tools(self):
        return [ToolDefinition(
            name="django-check",
            description="Check Django best practices",
            schema={},
            handler=lambda args: ToolResult(tool="django-check", content="OK"),
            permission="read",
        )]

---

Supported Providers

Provider	Models
OpenAI	GPT-4o, GPT-4, o1, o1-mini
Anthropic	Claude Sonnet, Claude Haiku
Google	Gemini Pro, Gemini Flash
Ollama	Local models (Llama, Mistral, etc.)

---

Commands

Command	Description
/connect	Connect to AI provider
/model	Select model
/project	Set project directory
/memory	Manage persistent memory
/export	Export generated prompt
/run	Execute shell command
/read	Read a file
/ls	List directory

---

Configuration

Via Commands

devobin
/connect        # Set up provider
/model gpt-4o   # Choose model

Via Config File

Edit ~/.config/devobin/settings.json:

{
  "provider": "openai",
  "model": "gpt-4o",
  "optimization_mode": "quick",
  "max_prompt_tokens": 8000,
  "memory_enabled": true
}

Environment Variables

export OPENAI_API_KEY="sk-..."
# or
export ANTHROPIC_API_KEY="sk-ant-..."

---

Architecture

User Input
    ↓
Scanner → Analyzer → Researcher
    ↓
Knowledge Graph
    ↓
Context Selector (intent + budget)
    ↓
Prompt Assembler (20 sections)
    ↓
Prompt Optimizer (optional)
    ↓
Final Prompt (.md)

---

Development

Setup

git clone https://github.com/mobinhasanghasemi/devobin.git
cd devobin
pip install -e ".[dev]"

Run Tests

pytest tests/ -v

Lint

ruff check devobin/

---

Project Structure

devobin/
├── runtime.py          # Main orchestrator
├── event_bus.py        # Event system
├── provider_router.py  # Smart provider routing
├── scanner/            # Workspace scanning
├── analyzer/           # Input analysis
├── researcher/         # Tech research
├── knowledge/          # Knowledge graph
├── context/            # Context selection
├── prompt/             # Assembly & optimization
├── tools/              # Tool system
├── plugins/            # Plugin system
├── config/             # Configuration
├── providers/          # LLM providers
├── cli/                # CLI interface
├── ui/                 # Textual UI
└── storage/            # Database

---

Links

• Website: devobin.ir
• PyPI: pypi.org/project/devobin
• GitHub: github.com/mobinhasanghasemi/devobin
• Instagram: devobin.ir
• Issues: GitHub Issues

---

License

MIT License — see LICENSE for details.

---

Author

Mobin Hasanghasemi

• GitHub: @mobinhasanghasemi
• Email: mobin.hasanghasemi.m@gmail.com