Zero to Hundred
Back to Marketplace

Mahakala Ai

Native local-first sovereign desktop AI powered by a custom C++ engine and glassmorphic UI.

Mahakala Ai

Type

Full Product

Tech stack

StripeCloudflareNode.jsPythonDocker

Documentation

Basic

A README with setup and run steps.

Maintenance

None

Sold as-is. No further updates from the Builder.

Description

Mahakala Local-First Sovereign AI Workspace

Mahakala is a high-performance, local-first AI workspace engineered around a native C++20 GGUF inference engine and a fluid Flutter client. Designed for complete data sovereignty and raw hardware execution, it includes an optional OpenAI-compatible HTTP server and a Node workspace sync service for flexible deployment across desktop, mobile, and web.

Core Local Execution Contract

  • Local-By-Default Execution: Daily workflows stay strictly on your machine. The internal LocalFirstPolicy router determines task execution boundaries—never the model.
  • Sanitized Cloud Escalation: Heavy tasks (extended context, intensive tool calls, or explicit deep reasoning) send at most a single sanitized payload (max 32 KB, keeping newest messages). Local file paths and credentials are automatically stripped, while thinking logs and system metadata remain strictly local.
  • Ephemeral Memory & Verification: Outbound payloads are purged upon response return with store: false. Every transaction generates a receipt recording the local wipe and exact byte counts.
  • Fail-Closed Privacy & Zero Bloat: Payloads matching path/credential patterns fail closed and will not send. Remote features require an explicit API token and HTTPS CORS origin (PRIVACY.md). The standard build is a pure coding agent—free of telemetry, wallets, marketplaces, or ads.

Architecture & Layout

  • packages/core-engine/ — Native C++20 inference engine, Dart FFI bindings, and HTTP server.
  • flutter_app/ — Cross-platform desktop, web, and mobile user interfaces.
  • server/ — Node.js service for shared-workspace synchronization.
  • docker/ — Containerized runtime with SHA-256-verified model fetching.
  • docs/ & scripts/ — Production setup matrices, architecture docs, and build harnesses.

Build & Runtime Requirements

  • Prerequisites: Flutter 3.x (.fvm/fvm_config.json), CMake $\ge$ 3.20, C++20 toolchain, Node 20+.
  • Packaging: Native execution via ./build.sh linux|windows|web (--local flag disables remote backends on web).
  • Model Execution: Point directly to any local GGUF file. Zero external downloads required unless explicitly initiated.

FAQ

What accounts or services do I need to set up before I can run this?
The default build is local-only and needs nothing external. If you want to use optional remote features, you will need to provide your own OpenAI-compatible endpoint or configure a Cloudflare worker. Stripe is included in the codebase but is not part of the default build.
What do I need to build this?
Flutter 3.x, CMake 3.20 or later, a C++20 toolchain, and Node 20+. The build script handles Linux, Windows, and web targets. For web, you need an HTTPS backend unless you build with the `--local` flag to disable remote features.
How do I get started?
Run `./build.sh` for your platform. Point the app at a GGUF model file you already have. Nothing is downloaded automatically. The local-first policy keeps your work on your machine by default.
What is included in the documentation?
Basic documentation covers architecture and production setup. The codebase includes a release matrix and internal competitive notes, but no user-facing guides beyond the README and PRIVACY.md.

Domains

aiproductivitydeveloper-tools

Seller

@mahakalaai-fqqn
@mahakalaai-fqqn
0.01 listings

Reviews

No reviews yet.

Mahakala Ai
$3,500

Share Listing

$3,500

Stats

Purchases0
Views5
Saves0
ListedOct 9, 2026

Trust signals

Government ID