What is measured
Status
Numbers, not labels. Vedākṣha publishes the measurements a reviewer would otherwise have to run themselves — version, residuals, test counts, code posture, deployment evidence. Every figure on this page is grep-able from the source tree or the linked package registry. No claims that aren’t.
Release
Current version
- Version
- 2.3.0
- Released
- 2026-04-29
- License
- BSL 1.1 (Apache 2.0 in 4 years)
- Toolchain
- Rust Edition 2024 · MSRV 1.85
Distribution
Where it ships
First-upload dates are stated explicitly. Vedākṣha is a young package on these registries; the engineering it represents is older, but the public distribution is dated from when it was first published.
| Channel | Package | Version | Note |
|---|---|---|---|
| crates.io | vedaksha (umbrella) + 8 sub-crates | 2.3.0 | First published 2026-04-24 |
| PyPI | vedaksha | 2.3.0 | First release (0.1.0) uploaded 2026-04-11 |
| GitHub Container Registry | ghcr.io/arthiqlabs/vedaksha-mcp | 2.3.0 | MCP server, stdio + HTTP transport |
| WASM | vedaksha-wasm | 2.3.0 | Browser / edge / Cloudflare Workers |
Validation
Measured against external references
The accuracy table below traces directly to the oracle JSON files committed under tests/oracle_jpl/. Each residual is recomputable from a fresh cargo test --workspace run.
- Oracle reference rows checked
- 8,700 across 3 JSON files in tests/oracle_jpl/
- Independent reference ephemerides
- Two — JPL Horizons plus a second reputable source
- Moon longitude residual (1900–2100, vs JPL Horizons)
- 0.23″ avg · 0.60″ max
- Planetary longitude — SpkReader (JPL DE440s)
- Sub-arcsecond, avg 1.7″
- Planetary longitude — AnalyticalProvider (VSOP87A + ELP/MPP02)
- <15″, avg 3.8″
- House cusps (10 systems)
- Sub-0.001° (SpkReader), sub-0.01° (AnalyticalProvider)
Tests & CI
Suite size and integration status
- Passing tests
- 832 across 9 test binaries
- Measured 2026-05-01 via
cargo test --workspace
Code posture
What the source enforces
- Workspace crates
- 10
unsafekeyword usage- 0 — every crate compiles without
unsafeRust - Crates with explicit
#![deny(unsafe_code)] - 8 of 10
- Edition
- 2024 across the workspace
- GitHub Actions workflows
- ci, audit, benchmarks, release
Production deployment
Where the engine is running
Live · agentic AI surface
KundaliMCP
The first deployed application built on Vedākṣha. Every chart, dasha, and yoga query at kundalimcp.com is computed by the same Rust crates published to crates.io as vedaksha. If the API is up, the engine is exercised.
Verify it yourself
Reproduce every number on this page
All claims above are derived from the public source tree. Clone, build, run the suite — the residuals, test counts, and code-posture facts are recomputable on your machine.
git clone https://github.com/arthiqlabs/vedaksha.git
cd vedaksha
# Test suite (832 passing tests as of v2.3.0)
cargo test --workspace
# Format + lint gates that CI enforces
cargo fmt --all -- --check
cargo clippy --workspace -- -D warnings
# Oracle row count (verifies the 8,700 figure)
python3 -c "import json; print(sum(
len(v) if isinstance(v, list) else sum(len(x) for x in v.values() if isinstance(x, list))
for f in ['comprehensive_reference.json', 'extended_reference.json', 'reference_positions.json']
for v in [json.load(open('tests/oracle_jpl/' + f))]
))"
# unsafe-code count (expect: 0)
grep -rE '\bunsafe\b' --include='*.rs' crates/ \
| grep -vE 'deny.unsafe_code|//.+unsafe' \
| wc -l