# Prism Answers API (https://www.socialcrawl.dev/platforms/prism/answers)

> Fans your question to multiple AI engines in parallel and keeps each answer verbatim, merges + dedupes their citations by normalized URL (with a `cited_by` list), and computes where they agree and disagree: a pairwise `agreement_matrix` + `disputed_claims` (the one labeled LLM step — degrades to null on failure) plus a deterministic `citation_overlap` (Jaccard, hand-verifiable). The second-opinion API for agents — one question, one bounded-cost call. Streams each engine's answer as it settles (SSE). `include=polymarket` adds probabilistic-market grounding. 15cr flat; a strict-majority engine failure refunds half (the raw answers are always the floor value).

TL;DR: `GET /v1/prism/answers` costs 15 credits per call and returns SocialCrawl's unified JSON schema. Single x-api-key auth, 100 free credits on signup.

## Parameters

| Parameter | Required | Description |
| --- | --- | --- |
| `query` | yes | The question, forwarded verbatim to every engine. |
| `engines` | no | CSV subset of perplexity,grok,tavily (default all three). |
| `include` | no | CSV of optional grounding legs: `polymarket` adds market probabilities (trimmed). |

## Code example

```bash
curl "https://www.socialcrawl.dev/v1/prism/answers?query=will%20the%20fed%20cut%20rates%20in%20september" \
  -H "x-api-key: sc_YOUR_API_KEY"
```

## FAQ

### What is the SocialCrawl AI answer API?

GET /v1/prism/answers is an AI answer API that sends one question to Perplexity, Grok, and Tavily in parallel, keeps each engine's answer verbatim, and merges their citations with an agreement check so you see where the engines align or disagree.

### Which engines does the answers API query?

Three: Perplexity, Grok (xAI), and Tavily. Pass engines= to narrow the fan-out (for example engines=perplexity,tavily), and include=polymarket to add prediction-market grounding to the answer.

### Does it return citations I can trace?

Yes. Every source is merged and deduped by URL with a cited_by list showing which engines used it. An agreement_matrix and disputed_claims highlight exactly where the engines converge or conflict.

### How much does an answer cost?

A flat 15 credits per call, regardless of how many engines fire. If a majority of engines fail the call partial-refunds, and if every engine fails the credits are refunded automatically.

### Can I stream the answer as it's generated?

Yes — it always streams over SSE. Each engine's answer arrives as it settles, then the merged citations, then a terminal done event, so agents and UIs show progress instead of waiting for the slowest engine.

See the full Prism API: https://www.socialcrawl.dev/platforms/prism

## Pricing

- Standard endpoints: 1 credit per call
- Advanced endpoints: 5 credits per call
- Premium endpoints: 10 credits per call
- 100 free credits on signup, no credit card required. Cached responses cost 0 credits. Credit packs never expire.
- Full pricing: https://www.socialcrawl.dev/pricing

## Explore with AI

Questions this API answers, phrased for an AI assistant:

- How do I get one answer from Perplexity, Grok, and Tavily at once?
- What does SocialCrawl's prism/answers endpoint return?