Documentation

Reference for the AstroGuru research platform: API, MCP, validation methodology.

Getting started

1. Visit calculate.astroguru.cc, sign in with Google.
2. Compute a sample chart from birth data.
3. Explore the research platform at app.astroguru.cc.
4. For programmatic access, see the REST API at api.astroguru.cc.

REST API

OpenAPI / Swagger UI is rendered at api.astroguru.cc/docs. Auth is Google ID token Bearer.

Endpoints

• GET /api/v1/health public liveness
• GET /api/v1/corpus/stats public corpus counts
• GET /api/v1/me authed user info
• POST /api/v1/charts/compute compute a Vedic chart (rate-limited per tier)
• POST /api/v1/validation/run run a hypothesis
• GET /api/v1/billing/status + POST /billing/checkout|portal|webhook Stripe subscription management
• GET /api/v1/admin/* admin endpoints (allowlist-gated)

MCP server

Model Context Protocol server at https://mcp.astroguru.cc/mcp?token=cna-mcp-.... Configure in Claude Desktop:

{
  "mcpServers": {
    "astroguru": {
      "type": "streamable-http",
      "url": "https://mcp.astroguru.cc/mcp?token=cna-mcp-..."
    }
  }
}

Tools: compute_chart, get_corpus_stats, list_public_hypotheses, run_hypothesis, get_panchanga.

Request a token by contacting nuwan@aigniteconsulting.ai.

Validation methodology

A hypothesis is two predicates: a chart-side predicate (e.g. Saturn in 8th in D1) and an outcome predicate (e.g. any death event). The validation engine computes four counts against the corpus:

• Matched. Charts whose chart-side predicate is true.
• Outcome. Charts with at least one event matching the outcome predicate.
• True positives. Charts in both sets.
• Total corpus. All computed charts (basis for precision/recall denominators).

From these we derive precision, recall, false positives, false negatives, true negatives.

Predicate types

Type	Required fields
planet_in_house	varga, planet, house
planet_in_sign	varga, planet, sign
planet_in_nakshatra	varga, planet, nakshatra
conjunction	varga, planet_a, planet_b
ascendant_in_sign	varga, sign
ascendant_in_nakshatra	varga, nakshatra
and / or / not	children (or child for not)
event_type / event_category	value (outcome side)
event_after_age / event_before_age	value (outcome side)

Statistical methods

Wilson score interval

Asymmetric, finite-sample-correct 95% confidence interval on precision (a binomial proportion). Default for all sample sizes. Wikipedia: Wilson score interval.

Beta-Binomial posterior

For small samples (matched count < 100), a Beta-Binomial posterior is reported in addition to Wilson. Uniform Beta(1, 1) prior; posterior is Beta(1 + TP, 1 + FP). Posterior mean and 95% credible interval reported.

Privacy and death-event policy

Aggregate stats are returned by default. Individual chart IDs only appear in the false-positive drill-down sample (up to 20 ids) and require explicit user action to reveal further detail.

Hypotheses involving death timing (event_type=cause_unspecified, event_category=death) require an explicit acknowledgement: death_event_ack: true in the validation request. Without it, the api returns 422 Unprocessable Entity. This is research-only territory.