AI-Native Architecture

Why Vedākṣha is Built for AI

Vedākṣha is the first astronomical computation platform designed from the ground up for AI agents. Every API decision, every data structure, every output format was chosen to make AI integration effortless and reliable.

Explore the AI Documentation

MCP Tool Catalog7 tools your agent can call Graph OutputCharts as property graphs Agent Patterns8 real-world workflows Why VedākṣhaFeature comparison matrix 5-Min QuickstartStart building in minutes

Design Pillars

10 reasons AI agents prefer Vedaksha

These are not afterthoughts bolted onto a legacy library. They are foundational decisions that shape every line of the codebase.

Pure Functions, Zero State

No initialization, no cleanup, no global state. Pass inputs, get outputs. Every computation is a pure function call — ideal for stateless AI agent invocations.

Semantic Type System

Body::Jupiter, not integer 5. Sign::Aries, not 0. Your agent reads typed enums that carry meaning, eliminating an entire class of mapping errors.

Graph-Native Output

Every chart is a ChartGraph with typed nodes and edges. AI agents excel at traversing structured relationships — and that is exactly what a chart becomes.

MCP-Native Protocol

7 tools exposed via the Model Context Protocol. OAuth 2.1 authentication, JSON-RPC 2.0 transport, streaming support. Your agent connects and starts computing.

Chart Highlights

Not every aspect matters equally. Vedaksha ranks significant chart features by strength and relevance, so your agent can summarize what matters most.

Natural Language Fields

Every transit event, yoga, and dasha period includes an nl_description field — a pre-written natural language explanation ready for your agent to relay.

Embedding-Ready Text

The EmbeddingTextEmitter produces optimized text chunks for vector stores. Build RAG pipelines over astrological knowledge without custom text extraction.

Streaming Results

Transit searches return Stream<TransitEvent> via MCP streaming. Your agent can process results as they arrive, not after the full search completes.

PII-Blind

The computation engine never sees personal data. It receives a Julian Day and coordinates — no names, no birth certificates, no data to protect.

Deterministic IDs

Same input always produces the same graph node and edge IDs. Your agent can compare charts across sessions, build incremental knowledge graphs, and detect duplicates.

The Contract

What your agent can rely on

Every MCP tool has a complete JSON schema — your agent knows the input format before calling.

Every error includes a structured error code and a self-correction hint — your agent can retry intelligently.

Every output field uses semantic types — your agent never needs to decode magic numbers.

Every computation is deterministic — same inputs always produce the same outputs.

Every chart can be emitted as a graph — your agent can store, query, and traverse relationships.

No computation requires prior state — your agent can call any tool at any time.

Start building.

Connect your AI agent to Vedākṣha in under 5 minutes.

Quickstart Guide MCP Tool Catalog