The classical corpus is in the engine, cited rule by rule. Shipping a Jyotish feature does not mean staffing someone to check it.
7 languages generated from one ontology and validated at build time — not a translation pass bolted on afterwards.
Never logged, never in an error message, never in our database. The computed chart may be cached — encrypted under a key derived from your own API key, so it is ciphertext even to us, and one parameter turns it off.
Output your agent uses as-is.
A headline it can quote, facts it can reason over, and pre-rendered prose it can pass through. No parsing. No schema mapping. No post-processing.
9 tools. No SDK.
Nothing to maintain.
Works with Claude, ChatGPT, Gemini, or any MCP-compatible agent.
Defend every reading you ship.
Call pramaan on any claim in a chart result, addressed by the concept URI the response already gave you. KundaliMCP traces it from the VSOP87A sky position and the ayanamsha, through chart casting, to the classical verse. No black boxes. Ever.
Seven languages, without a translation vendor.
Anuvada is not a translation layer bolted on top. Every Vedic concept — yoga, dosha, nakshatra, graha — is a JSON-LD entity with axes, grammar rules, and validated surfaces in all seven languages from compile time. Add a new language without touching the engine.
anuvada/ directory. The engine compiles it automatically — no code changes required.No third party in
your critical path.
No third-party ephemeris. No vendor lock. The entire stack from raw sky coordinates to MCP response is owned and auditable. A new primitive flows sky to screen in days.
Connect in 60 seconds.
Three steps. No library to install. No adapter layer to maintain.
Want the whole contract in one file? Full tool reference (.md) →Start free. Scale when you need.
No per-seat fees. No feature gates. Usage-based.
All tiers · birth inputs never logged · AES-256-GCM caller-keyed cache · OAuth 2.1 · clean-room Rust ephemeris