⚡ DevSense — Autonomous AI for Your Codebase

Understand. Build. Fix. Ship. > AI-powered terminal-native coding intelligence for developers who live in the CLI.

DevSense is an open-source AI coding system that helps you:

• Understand unfamiliar repositories instantly
• Explain files, architecture, and dependencies
• Chat with your codebase directly from the terminal
• Toggle between Cloud (Gemini/Mistral) or Local (Ollama) AI models
• Scale toward fully autonomous coding workflows

👉 No IDE plugin | 👉 No editor lock-in | 👉 No context switching

🌐 Website: DEVSENSE

---

🧠 What DevSense Can Do

🔎 Understand Any Codebase Instantly

Get architecture-aware answers using real repository context.

devsense ask "How does authentication work?"

📄 Explain Complex Files in Seconds

Understand purpose, dependencies, and logic without manual scrolling.

devsense explain src/auth.ts

🗺️ Map Your Project Architecture

Quickly locate core modules like Auth, API, Database, and Services.

devsense map

💬 Chat With Your Repository

Interactive AI-powered terminal chat built around your specific codebase.

devsense chat

---

⚡ Multi-Provider AI Support

Choose the backend that fits your workflow:

Cloud Providers	Local Providers (via Ollama)
Gemini	Llama 3 / CodeLlama
Mistral	DeepSeek / Qwen
OpenAI (Planned)	Gemma / Mistral

Run Fully Local:

devsense init
# Select Ollama -> Choose llama3

---

🚀 Why DevSense?

Most AI coding tools are trapped inside editors. DevSense is built for Real Developers:

• ✅ Terminal-native: Stays in your flow.
• ✅ Repo-aware: Understands the "big picture."
• ✅ Privacy First: Support for 100% offline local models.
• ✅ Extensible: Open-source architecture ready for custom analyzers.
• ✅ Future-Proof: Moving toward an autonomous "Claude Code" alternative.

---

⚙️ System Architecture

User ➔ CLI Commands ➔ Project Scanner ➔ Analyzer ➔ Context Builder ➔ AI Provider Layer ➔ Response

---

🧩 Current Core Features

Feature	Status
Repo Scanning & File Explanation	✅ Complete
Architecture Mapping	✅ Complete
Gemini / Mistral Support	✅ Complete
Ollama Local AI Support	✅ Complete
Context DB (SQLite)	🚧 Phase 1
Tool Execution Layer	🚧 Planned
Autonomous Coding Loop	🚧 Planned

---

🚀 Quick Start

1. Install Globally

npm install -g devsense

2. Initialize Inside Your Project

devsense init

3. Configure Provider

• For Cloud: Add your key to .devsense/.env (e.g., GEMINI_API_KEY=your_key)
• For Local: Install Ollama and pull a model:

  ollama pull llama3

4. Start Using It

devsense summary
devsense ask "What does this project do?"
devsense explain src/server.ts
devsense chat

---

🛠️ Commands Reference

Command	Description
devsense init	Initialize project configuration
devsense summary	Get a high-level project overview
devsense ask "<q>"	Ask questions about the repo
devsense explain <f>	Detailed explanation of a specific file
devsense chat	Open interactive repo chat
devsense map	Visualize project architecture
devsense doctor	Diagnose setup and model availability

---

🗺️ Roadmap

• Phase 0 ✅ Local AI: Ollama integration and multi-backend support.
• Phase 1 🚧 Context DB: SQLite per-project memory and incremental scans.
• Phase 2 🚧 Tool Layer: Capability to read/write files and run terminal commands.
• Phase 3 🚧 Diff Editing: Safe patches and incremental code modifications.
• Phase 4 🚧 Autonomous Agent: Generate ➔ Execute ➔ Debug ➔ Retry loop.

---

🛠️ Build From Source

git clone https://github.com/your-repo/devsense
cd devsense
npm install
npm run build
npm link

---

🤝 Contributing & Support

We welcome feature requests, new provider integrations, and analyzer plugins!

1. Fork the repo.
2. Create your feature branch.
3. Submit a PR.

If DevSense helps you, please star the repository ⭐