Getting Started

Installation

# Install globally
npm install -g @zosmaai/dhara

# Or run directly with npx
npx @zosmaai/dhara "List files in this project"

Quick Start

One-shot mode

Pass a prompt as an argument:

dhara "What files are in this project?"

Interactive mode

Run without arguments to enter the TUI (full-screen terminal UI):

dhara

Or use the line-based REPL:

dhara --repl

Configure a Provider

Dhara needs an LLM provider to function. Set the corresponding environment variable and use --provider:

export OPENAI_API_KEY="sk-..."
dhara --provider openai --model gpt-4o

export ANTHROPIC_API_KEY="sk-ant-..."
dhara --provider anthropic --model claude-sonnet-4-20250514

export GOOGLE_API_KEY="..."
dhara --provider google --model gemini-2.5-flash

export MISTRAL_API_KEY="..."
dhara --provider mistral --model mistral-large-latest

export GROQ_API_KEY="gsk_..."
dhara --provider groq --model llama-3.3-70b-versatile

Built-in providers (no extra deps)

# OpenAI (also works with OpenAI-compatible endpoints)
export OPENAI_API_KEY="sk-..."
dhara --provider openai --model gpt-4o

# Anthropic
export ANTHROPIC_API_KEY="sk-ant-..."
dhara --provider anthropic

# Opencode-go (default)
export OPENCODE_API_KEY="..."
dhara

pi-ai providers (auto-detected)

All pi-ai providers use their standard environment variables:

Provider	Env Var	Example
Google	`GOOGLE_API_KEY`	`dhara --provider google --model gemini-2.5-flash`
Mistral	`MISTRAL_API_KEY`	`dhara --provider mistral`
Groq	`GROQ_API_KEY`	`dhara --provider groq`
DeepSeek	`DEEPSEEK_API_KEY`	`dhara --provider deepseek`
Amazon Bedrock	AWS credentials	`dhara --provider amazon-bedrock`
Azure OpenAI	`AZURE_OPENAI_API_KEY`	`dhara --provider azure-openai-responses`
Fireworks	`FIREWORKS_API_KEY`	`dhara --provider fireworks`
OpenRouter	`OPENROUTER_API_KEY`	`dhara --provider openrouter`

Project Configuration

Create a .dhara/settings.json in your project root:

{
  "provider": "google",
  "model": "gemini-2.5-flash",
  "maxIterations": 15,
  "autoSave": true
}

Context Files

Place AGENTS.md or CLAUDE.md in your project root. Dhara walks up from the working directory to find them.

# AGENTS.md

- Run `npm run check` before committing
- Keep responses concise
- Never modify production data

CLI Reference

dhara <prompt> [options]     One-shot mode
dhara [options]              TUI mode (default)
dhara --repl [options]       REPL mode
dhara session list           List saved sessions
dhara session delete <id>    Delete a session
dhara session info <id>      Show session details

Options:
  --provider <name>     LLM provider
  --model <id>          Model ID
  --base-url <url>      Custom API base URL
  --cwd <path>          Working directory
  --resume <id>         Resume a session
  --json                JSON output (for CI/CD)
  --theme <name|path>   TUI theme
  --repl                Use REPL instead of TUI
  --version             Show version
  --help                Show help

Getting Started

On this page