Delena-AI

Delena-AI is a read-only CLI tool for loading and exploring codebases using AI. You can point it at local directories or remote GitHub repositories and ask questions about the code — either interactively or via a one-shot command suitable for automation or agent integration. (GitHub)

---

🚀 Features

Delena-AI provides:

✔️ Load local directories or clone remote repositories

✔️ Read-only AI exploration — no modifications to source code (GitHub)

✔️ Session management (save, resume)

✔️ Multiple AI models supported

✔️ Token counting per query

✔️ One-shot CLI mode for automation and agents

✔️ JSON output for parsing in scripts or tools (GitHub)

---

📦 Installation

Clone the repo and install dependencies with Bun:

git clone https://github.com/support-agent-org/Delena-AI.git
cd Delena-AI
bun install

Create a .env file with your AI API key:

GOOGLE_GENERATIVE_AI_API_KEY=your_api_key_here

---

💬 Usage

🔹 Interactive Mode

Start the interactive CLI:

bun run src/index.ts

Inside the interactive prompt, you can:

/load ./my-project
What are the tests in this repo?
Explain the architecture

---

🔹 One-Shot Mode (Agent-Friendly)

Run a single question from the shell — ideal for scripts and automation:

bun run query --repo <path|url> "Your question here"

---

⚙️ CLI Options

Flag	Description
-r, --repo <path>	Local directory or Git URL (required)
-j, --json	Output structured JSON
-m, --model <model>	AI model to use (default: google/gemini-3.0-flash)
-q, --quiet	Output only the answer (suppress logs)	(GitHub)

---

📊 JSON Output Format

When using --json, responses are formatted for easy machine parsing:

{
  "success": true,
  "answer": "This project is ...",
  "repository": "my-project",
  "model": "google/gemini-3.0-flash",
  "tokens": {
    "input": 1234,
    "output": 567,
    "total": 1801
  }
}

If the query fails:

{
  "success": false,
  "error": "Failed to load repository"
}
``` 

---

## 🧠 Interactive Commands

Once inside the CLI, use:

/load <path|url> # Load a project /unload # Unload current repo /status # View current context and tokens /save # Save session /resume # Resume saved session /sessions # List saved sessions /model # Change model/provider /mode <low|medium|high> # Response depth /help # List commands /exit # Quit

---

## 🗄️ Session Storage

Delena-AI stores sessions and cached clones in:

~/.support-agent/

- `sessions.json`: session metadata  
- `repos/`: cached cloned repositories 

---

## 🛡️ Read-Only Policy

Delena-AI **cannot modify code**. It can:

✔️ Read files & directories  
✔️ Clone repositories (shallow)  
✔️ Search files using glob  
✔️ Fetch external URLs  

Delena-AI **cannot**:

❌ Write, delete, or modify files  
❌ Execute system commands 

---

## ⚙️ Configuration

Customize behavior using `opencode.json` settings. Example:

```json
{
  "agent": {
    "support": {
      "model": "google/gemini-3-pro-preview",
      "tools": {
        "read": true,
        "glob": true,
        "fetch": true,
        "write": false,
        "bash": false
      }
    }
  }
}
``` 

---

## 📈 Supported Models & Providers

Delena-AI works with multiple AI backends and models — set via CLI or config.  
Default is `google/gemini-3.0-flash`, but you can switch to other supported models. 

---

## 🤝 Contributing

Contributions are welcome! To contribute:

1. Fork the repository  
2. Create a branch (`git checkout -b feature/xyz`)  
3. Commit your changes  
4. Push to your fork  
5. Open a Pull Request

---

## 📄 License

This project is released under the **MIT License**.