KundaliMCP — first product on Vedākṣha is now live.Visit →
Vedākṣha

Why Vedaksha

The ephemeris engine should have existed.

We built Vedaksha because the astronomical computation industry has a set of structural problems that no amount of wrapper libraries can fix. The engine itself needs to be redesigned — for modern platforms, for Vedic traditions, for AI agents, and for developers who ship commercial products.

01

Copyleft Licensing

The Problem

Most ephemeris engines use AGPL or GPL. If you build a commercial product on top, you must disclose your entire source code. This forces teams to either open-source their app, pay license fees, or build risky workarounds.

Vedaksha

Vedaksha uses BSL 1.1 — free for non-commercial use, $500 one-time for commercial. No source disclosure. No copyleft. Converts to Apache 2.0 after five years.

02

Vedic as an Afterthought

The Problem

Existing computation engines were designed for Western tropical astrology in the 1990s. Vedic features — nakshatras, dashas, vargas, yogas, Shadbala — are bolted on through wrapper libraries maintained by third parties, with no guarantee of correctness.

Vedaksha

Vedaksha is Vedic-first. 5 dasha systems, complete 5-limb panchanga, 27 nakshatras with deity/yoni/nadi, graded drishti, degree-precise Shadbala, 16 vargas with school-variant support, 44 ayanamsha systems. All in the core type system, cited to BPHS.

03

C-Only, Desktop-Era Architecture

The Problem

Legacy engines are written in C, designed for desktop GUI applications. Integrating them into modern stacks — mobile apps, web browsers, serverless functions, edge compute — requires FFI bindings, build system complexity, and platform-specific compilation.

Vedaksha

Vedaksha is pure Rust. Compiles natively, to WebAssembly (972 KB, zero data files), and to Python via PyO3. Runs in browsers, Cloudflare Workers, AWS Lambda, Docker, and bare metal. One codebase, every target.

04

Not Built for AI

The Problem

AI agents need typed, structured, queryable output. Legacy engines return arrays of floating-point numbers. Making them useful for an LLM requires building a translation layer — parsing output, adding context, structuring responses.

Vedaksha

Vedaksha produces property graphs with 10 typed node types and 13 edge types. Emit directly to Neo4j Cypher, SurrealDB, JSON-LD, or RAG-optimized text. The MCP server exposes 7 typed tools with JSON schemas — any MCP-compatible agent can call them without custom prompting.

05

Unverifiable Accuracy

The Problem

Most engines derive their algorithms from other software rather than from primary published sources. This creates a chain of implementation copies where errors propagate and correctness cannot be independently verified.

Vedaksha

Vedaksha is a clean-room implementation. Every algorithm traces to a published NASA, IAU, or academic source — Meeus, Chapront, Bretagnon, BPHS, Phaladipika. Planetary positions are validated against JPL Horizons DE441. The osculating node achieves <0.03° accuracy vs JPL. 528 tests with 24,000+ oracle validation data points.

06

No Privacy Model

The Problem

Ephemeris computation inherently processes birth data — date, time, location. Most engines have no concept of data classification, making GDPR and privacy compliance an application-level burden.

Vedaksha

Vedaksha is PII-blind by design. The engine accepts Julian Day and geographic coordinates — no names, no dates in human-readable form. The graph output carries a DataClassification tag (Anonymous, Pseudonymized, Identified) so downstream systems can enforce retention and access policies.

By the Numbers

528+

Automated tests

24K+

Oracle validation points

<0.03°

Node accuracy vs JPL DE441

44

Ayanamsha systems

5

Dasha systems

50

Vedic yogas detected

7

Languages (localization)

972 KB

WASM binary (zero data files)

Ready to try it?

Install from crates.io, pip, or npm. Run the MCP server locally or via Docker. Free for non-commercial use.