✦ Open source · Fully local · No API key

AI coding assistant
powered by Ollama

Reads, edits, and creates files in any project. Runs fully local with any Ollama model — private, fast, no API key required.

qwen2.5-coder deepseek-r1 llama3.2 mistral gemma3

$ curl -fsSL https://manthra.informaticsint.au/install | bash

↓ Download manthra-linux-x64

chmod +x manthra-linux-x64 && mv manthra-linux-x64 ~/.local/bin/manthra

↓ Download manthra-linux-arm64

chmod +x manthra-linux-arm64 && mv manthra-linux-arm64 ~/.local/bin/manthra

↓ Download manthra-macos-x64

chmod +x manthra-macos-x64 && mv manthra-macos-x64 ~/.local/bin/manthra

⚠ macOS will block this binary — follow these steps

macOS blocks unsigned binaries. After downloading, run these commands in Terminal:

xattr -c ~/Downloads/manthra-macos-x64
chmod +x ~/Downloads/manthra-macos-x64
codesign --force --deep --sign - ~/Downloads/manthra-macos-x64
mv ~/Downloads/manthra-macos-x64 ~/.local/bin/manthra

── or go to System Settings → Privacy & Security → "Allow Anyway" ──

💡 Easiest fix: use the curl installer above — it handles this automatically.

↓ Download manthra-macos-arm64

See steps below — macOS requires signing before running.

⚠ macOS will block this binary — follow these steps

macOS blocks unsigned binaries. After downloading, run these commands in Terminal:

xattr -c ~/Downloads/manthra-macos-arm64
chmod +x ~/Downloads/manthra-macos-arm64
codesign --force --deep --sign - ~/Downloads/manthra-macos-arm64
mv ~/Downloads/manthra-macos-arm64 ~/.local/bin/manthra

── or go to System Settings → Privacy & Security → "Allow Anyway" ──

💡 Easiest fix: use the curl installer above — it handles this automatically.

↓ Download manthra-win-x64.exe

Place in a folder on your PATH, e.g. C:\Users\you\bin\manthra.exe

Or install via Git Bash (recommended):

curl -fsSL https://manthra.informaticsint.au/install | bash

Or via WSL2 (Ubuntu/Debian terminal):

curl -fsSL https://manthra.informaticsint.au/install | bash

manthra — ~/my-project

✦ MANTHRA.md loaded

refactor the auth module to use JWT instead of sessions

⠸ Analyzing…

✓ read src/auth/session.ts

import { SessionStore } from './store'

export async function createSession(userId: string) {…}

✓ read src/auth/middleware.ts

✓ bash npm list jsonwebtoken

I'll refactor the auth module to use JWT. Here's the plan:

1. Replace SessionStore with jsonwebtoken

2. Update createSession → signToken

3. Update middleware to verify JWT on each request

✓ edit src/auth/session.ts

✓ edit src/auth/middleware.ts

✓ Done. Auth module now uses JWT. Tests still pass.

■ Chat · Ollama · qwen2.5-coder:latest · 12s

Features

Everything your AI needs
to work with your code

No cloud sync, no IDE plugin. Just a fast CLI that does the work.

🟣

Ollama-native AI

Runs with any model available in your local Ollama install — qwen2.5-coder, deepseek-r1, llama3, mistral, and more. No API key ever.

🔧

Full tool use

Reads, writes, and edits files. Runs shell commands. Fetches URLs. Makes HTTP requests. The AI does real work, not just advice.

📄

MANTHRA.md project context

Drop a MANTHRA.md in your repo and Manthra loads it automatically for any AI provider. Generate one with /init.

🔒

CWD-aware permissions

File operations within your project are always allowed. Paths outside your working directory prompt for confirmation before proceeding.

⚡

Streaming responses

Live streaming with a fixed terminal layout — status bar and input stay pinned at the bottom while responses scroll above.

🧠

Persistent memory

Manthra remembers context across sessions. Configure providers and models through the built-in web UI with manthra web.

Get started

Up and running in 60 seconds

Install with one command

The installer detects your OS and architecture, downloads the right binary to ~/.local/bin/manthra, and adds it to your PATH.

curl -fsSL https://manthra.informaticsint.au/install | bash

Configure Ollama

Open the web UI to add your local Ollama instance, pick a model, and test the connection. No API key needed.

manthra web

Run it in any project

Navigate to your project directory and launch Manthra. It reads your files relative to where you launched it.

cd ~/my-project && manthra

Add project context (optional)

Generate a MANTHRA.md file so the AI instantly understands your project conventions, structure, and how to build it.

/init

Running locally with Ollama

Use Manthra completely offline — no API key needed

Ollama lets you run open-source models (Llama, Mistral, Gemma, DeepSeek, and more) entirely on your own machine. Free, private, no rate limits.

Step 1 — Install Ollama

Download from ollama.ai or install via terminal:

          # macOS / Linux

          curl -fsSL https://ollama.ai/install.sh | sh

Windows: download the installer from ollama.ai

Step 2 — Pull a model

Pull any model from the Ollama library. Recommended for coding:

          ollama pull llama3.2

          # or larger models:

          ollama pull qwen2.5-coder:7b

          ollama pull deepseek-r1:7b

Step 3 — Start Ollama

Start the Ollama server (it runs on 127.0.0.1:11434):

          ollama serve

          # macOS: Ollama starts automatically

          # after installation

Step 4 — Connect in Manthra

Open the Manthra web UI, add an Ollama instance, and click List Models to see your local models:

          manthra web

          # → Add Instance

          # Base URL: http://127.0.0.1:11434

          # → List Models → select one → Save

💡 Tip: Once Manthra is configured with Ollama, launch it from your project as normal — manthra. Everything works offline. Use manthra web anytime to switch models or add another Ollama instance.

Compatibility

Runs everywhere you code

Single binary — no Node.js, no runtime required.

Linux

x64 and arm64. Works on Ubuntu, Debian, Fedora, Arch, and any distro.

x64 arm64

macOS

Intel and Apple Silicon (M1/M2/M3). Install via curl or download directly.

⚠ Gatekeeper blocks unsigned binaries

After a manual download, open Terminal and run:

xattr -c ~/Downloads/manthra-macos-arm64
codesign --force --deep --sign - ~/Downloads/manthra-macos-arm64
chmod +x ~/Downloads/manthra-macos-arm64

✓ The curl installer does this automatically.

x64 (Intel) arm64 (M1/M2/M3)

Windows

Three ways to run Manthra on Windows:

Git Bash (recommended)

curl -fsSL https://manthra.informaticsint.au/install | bash

WSL2 (Ubuntu/Debian)

curl -fsSL https://manthra.informaticsint.au/install | bash

Direct .exe download

↓ manthra-win-x64.exe

Add the folder to your PATH in System Properties

x64 Windows Terminal recommended

AI coding assistant
powered by Ollama

Everything your AI needs
to work with your code

Ollama-native AI

Full tool use

MANTHRA.md project context

CWD-aware permissions

Streaming responses

Persistent memory

Up and running in 60 seconds

Install with one command

Configure Ollama

Run it in any project

Add project context (optional)

Use Manthra completely offline — no API key needed

Any model in your Ollama library

Runs everywhere you code

Linux

macOS

Windows

Ready to ship faster?

AI coding assistantpowered by Ollama

Everything your AI needsto work with your code

Ollama-native AI

Full tool use

MANTHRA.md project context

CWD-aware permissions

Streaming responses

Persistent memory

Up and running in 60 seconds

Install with one command

Configure Ollama

Run it in any project

Add project context (optional)

Use Manthra completely offline — no API key needed

Any model in your Ollama library

Runs everywhere you code

Linux

macOS

Windows

Ready to ship faster?

AI coding assistant
powered by Ollama

Everything your AI needs
to work with your code