AI Interpretation API — Currently Disabled
The optional AI prose layer is currently disabled. Core calculation endpoints already include their versioned rule interpretation where applicable. The planned AI report rewrites selected existing facts and rules rather than replacing numerical calculation.
POST /v1/reports50,000 credits per monthly allowance · no trial · no automatic overage charges. Plan details ↗
What this API provides.
Review the optional fact-constrained AI report workflow and current ai_not_available response. Rule-based interpretation is already included in core APIs.
A completed result as the source, not an unconstrained image prompt
Selected facts and rules kept separate from authoritative numbers
Provider/model, prompt-schema and usage metadata when enabled
Send
- A completed source_operation_id owned by your organization
- format: ai and the source result locale
- Current public requests return ai_not_available before a model call
Receive
- Current behavior: HTTP 503 ai_not_available
- Intended result: a fact-constrained prose layer with provenance
- No AI response is represented as generated in this public sample
Current response example.
A local route check with synthetic authentication records the actual HTTP 503 rejection. The feature gate runs before metering or a provider call.
{
"source_operation_id": "1299830cdfe3ed0bb78549f7be780e77",
"format": "ai",
"locale": "en"
}{
"request_id": "9502e388d2624b18a79b7a6237046ab9",
"error": {
"code": "ai_not_available",
"message": "Ai not available",
"details": null
}
}Placeholder upload/source IDs are documentation inputs only. The route gate rejects this request before looking up any private object.
cURL request — use your own API key and a new idempotency key
curl --request POST 'https://api.celestackapi.com/v1/reports' \
+ --header 'Authorization: Bearer YOUR_API_KEY' \
+ --header 'Content-Type: application/json' \
+ --header 'Idempotency-Key: ai-interpretation-api-example-001' \
+ --data '{
"source_operation_id": "1299830cdfe3ed0bb78549f7be780e77",
"format": "ai",
"locale": "en"
}'Example computed 2026-11-17. Full fixture manifest, engine versions and SHA-256 checksums ↗
Ways to use it.
- Plan a separate prose layer on top of authoritative calculations
- Compare the included rule content before considering optional AI
- Review the gated workflow without paying for speculative generation
12 response locales
English, Ukrainian, Japanese, Spanish, French, German, Portuguese, Arabic, Italian, Hindi, Chinese and Korean.
Calculation values remain language-independent. Localized rule text is agent-authored; independent native-language review remains pending. Machine-readable field names stay stable.
Limits and interpretation.
- AI generation is disabled publicly. There is no generated-AI example or promise of availability.
- When enabled, AI prose will remain symbolic, and selected facts limit coverage. Numerical results must stay unchanged.
- The listed 500-credit rate is the configured successful-job price, not an additional charge for the included rule interpretation.
Interpretive content is for symbolic reflection and entertainment. It is not a verified personal prediction or professional advice. Read the API conventions ↗
Choose your billing.
Both options include 50,000 credits per monthly allowance, resetting monthly. No trial and no automatic overage. Custom needs use a contact form.
Saved results: 7-day retention, up to 128 MiB per organization on this initial shared deployment, with a 512 MiB shared retained-result limit. Capacity can return a temporary error; there are no automatic extra charges. Production artifacts have separate retention. All usage and resource limits ↗. Checkout availability is shown in your account.
Frequently asked questions.
Do I need AI for localized interpretation?
No. Supported core endpoints already include versioned rule-based interpretation in 12 locales, except the lower-cost positions endpoint which returns no interpretation.
Can I call the AI report now?
No. ai_enabled is false in the live catalogue and the route returns ai_not_available. No paid model call was made for the example.
Will AI calculate the planetary positions?
No. The calculation engine produces the authoritative data. The optional layer is designed to rewrite selected existing facts and rules, not invent positions or change numerical results.
What does access cost?
Access is USD 39.99 per month or USD 199 billed annually, with no trial. Both billing options provide 50,000 credits per monthly allowance. 500 credits per successful request or job. Daily credit, attempt, rate and resource limits apply. There are no automatic overage charges. This module is currently gated; a subscription does not enable it.