dwata

Name: dwata
Author: brainless

Verified

AI enabled insights from emails, calendars, contacts, files, Slack, databases, web... Fast, private and local. Launching soon!

179stars

13forks

Rust

Installation

# Add to your Claude Code skills
git clone https://github.com/brainless/dwata

Getting Started

Guides for using ai agents skills like dwata.

Caveman: Cut Claude Token Use by 65%
How agent-side prompt compression works, when to use it, and when not to.
What is an AI Skills Marketplace?
Definitions, how marketplaces work, and how to choose between them in 2026.
Getting Started with AI Skills

Security ReportVerified

Last scanned: 5/30/2026

{
  "issues": [],
  "status": "PASSED",
  "scannedAt": "2026-05-30T15:48:39.628Z",
  "npmAuditRan": true,
  "pipAuditRan": true
}

README.md

Frequently Asked Questions

What is dwata?

dwata is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by brainless. AI enabled insights from emails, calendars, contacts, files, Slack, databases, web... Fast, private and local. Launching soon!. It has 179 GitHub stars.

Is dwata safe to use?

Yes. dwata passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.

How do I install dwata?

Clone the repository with "git clone https://github.com/brainless/dwata" and add it to your Claude Code skills directory (see the Installation section above).

What programming language is dwata written in?

dwata is primarily written in Rust. It is open-source under brainless on GitHub, so you can review or fork the full source.

Are there alternatives to dwata?

Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh dwata against similar tools.

Agentic AI for Beginners

Build your first AI agent from scratch - tool use, ReAct pattern, memory, deployment

41 minBeginner

Comments (0)

to leave a comment.

No comments yet. Be the first to share your thoughts!

Related Skills

superpowers

by obra

An agentic skills framework & software development methodology that works.

234,966

timeline-studio zylos-core

Dwata

Connect your email inbox, download emails, and use LLM agents to extract financial transaction data — all running locally on your machine.

All data stays on your machine. dwata works with Ollama and local models (tested on Mac Mini M4 16GB), so your emails never leave your computer.

[!WARNING] dwata is very early software and is being developed actively, I am sorry if the extracted data has bugs.

What dwata does today

Email inbox

Connect your Gmail or IMAP account. dwata downloads your emails and stores them locally in SQLite.

Email inbox

Financial template detection

Select financial emails and run an LLM agent to generate extraction templates. The agent reads sample emails and produces reusable patterns — you only need AI once per email sender.

Detect financial templates

Running template detection with Ollama

Financial templates

Browse and manage the generated templates. Each template captures how to extract financial data from a specific sender.

Financial templates

Extracted transactions

Once templates are in place, dwata extracts financial transactions from matching emails automatically.

[!WARNING] There are quite a few issues with the extraction logic. I am working on it actively.

Financial transactions

Financial bills

LLM settings

Use Ollama with a local model (Ministral 3: 3b), OpenAI (GPT-4o Nano), or Google Gemini (Gemini 2.5 Flash Preview). Switch models in settings.

Settings - Ollama Ministral 3 3b

Privacy

All email data is stored locally in SQLite — never sent to a cloud service
With Ollama, template detection runs entirely on your machine with no external API calls
OS keychain stores credentials (Gmail password / OAuth tokens) — dwata uses a single keychain entry so you get one prompt on first launch
On macOS, select Always Allow when the keychain prompt appears so you're not asked again

Getting started

Download the latest release for your platform from GitHub Releases.

Launch the Dwata desktop app (Tauri). It starts the dwata-api sidecar automatically and loads the GUI inside the app window.

Connecting Gmail

dwata supports Gmail via OAuth. Set your Google OAuth client_id and client_secret in the config file:

macOS: ~/Library/Application Support/dwata/project.toml
Linux: ~/.config/dwata/project.toml
Windows: %APPDATA%\dwata\project.toml

You can use your own Google OAuth app (bring-your-own credentials).

Using Ollama

Install Ollama and pull model Ministral 3:3b:

ollama pull ministral-3:3b

Then set the model in dwata's settings page.

Tech stack

Backend: Rust + Actix-web
Database: SQLite (local)
Frontend: SolidJS
UI Components: DaisyUI
LLM: Ollama (local), OpenAI, or Google Gemini

Support

Bugs & issues: GitHub Issues
Discussions: GitHub Discussions
Developer guide: DEVELOP.md

License

GPL v3 — see LICENSE.

From the founder

I am Sumit, and I live in a small eastern Himalayan village in India. I mentor/co-mentor hundreds of folks each month about how to use coding agents. I run a digital nomad space in our village. Come, say Hi!