The grounded sports API
Receipt-backed, refusal-aware ratings across 5 sports and 65,554 players (1876–2026) — facts your app or AI agent can actually cite. Metered in StatVault Credits: pay only for what you call.
No credit card required for the free plan.
Live KC rollout is still in progress for developer API surfaces.
The API and widget documentation should stay explicit about rollout state. Until the live KC seed is complete, downstream consumers should expect honest fallback labeling rather than silent empty responses.
Developer-facing ritual contract
Integrators should treat the public API as one connected weekly ritual, not a pile of isolated endpoints. The movement surface, evidence surface, and commitment surface are all part of the same trust loop.
Public feature registry
Why KC Moved
Weekly command center for movers, evidence, operator demand, and next actions.
Beat the Vault
Make weekly KC calls, frame rivalries, and track whether your conviction was right.
Ask Knox
Natural-language query surface for current player and product questions.
Vantage Query
Fast, reusable Knox answer surface for common weekly questions.
Knox Cases
Durable evidence reviews captured from Ask Knox and Vantage Verify.
Start free — then buy credits, no subscription
Every account gets a free monthly credit grant. Need more? Buy a pack or turn on auto-refill below — you pay only for the calls you make.
- ✓ 1,000 credits / month, free — no card to start
- ✓ Every endpoint: rankings, valuation, grounded rating & AI answer
- ✓ Receipt-backed responses — every answer is verifiable
- ✓ Then buy packs or enable auto-refill — pay only for what you call
Priced in Credits — the same currency as every plan
A raw read is 1 credit; a grounded, receipt-backed AI answer is 25. Beyond your plan's monthly grant, top up with pay-as-you-go packs — the more you buy, the less each credit costs.
| Call type | Credits |
|---|---|
| Rankings / search / leaderboardsraw reads | 1 |
| Player valuation / comps / career-shapederived analytics | 2 |
| Grounded rating (receipt-backed, refusal-aware)citable, no LLM | 5 |
| Evidence ledger / receipt retrievalprovenance | 5 |
| Grounded AI answer (LLM synthesis)the premium unit | 40 |
| Bulk snapshot / export slicedata licensing | 500 |
Credit packs
- $10
10,000 credits
$1.00 / 1k
- $40
50,000 credits
$0.80 / 1k
- $150
250,000 credits
$0.60 / 1k
Overage beyond a plan is billed per credit — never a hard cutoff.
Quick Start
// A receipt-backed cross-sport rating your agent can cite.
const res = await fetch(
'https://statvault.org/api/v1/grounded/rating?sport=nfl&slug=barry-sanders',
{ headers: { 'X-API-Key': 'sv_live_YOUR_KEY' } }
);
const data = await res.json();
console.log(data.claim); // "barry sanders's best season (1990) ranks #… "
console.log(data.provenance); // method + artifact + attribution
// credits: res.headers.get('X-StatVault-Credits-Remaining')import httpx
r = httpx.get(
"https://statvault.org/api/v1/grounded/rating",
params={"sport": "nba", "slug": "lebron-james"},
headers={"X-API-Key": "sv_live_YOUR_KEY"},
)
data = r.json()
print(data["claim"]) # grounded, or {"grounded": false} on a refusal
print(r.headers["X-StatVault-Credit-Cost"]) # 25curl -s \
-H "X-API-Key: sv_live_YOUR_KEY" \
"https://statvault.org/api/v1/grounded/rating?sport=mlb&slug=babe-ruth" \
| jq '{claim, provenance, creditsRemaining}'Endpoints Reference
| Method | Endpoint |
|---|---|
| POST | /api/mcp |
| GET | /api/v1/grounded/rating |
| GET | /api/v1/rankings/:sport |
| GET | /api/v1/players/search |
| GET | /api/v1/vantage/value |
| GET | /api/v1/vantage/clone/:player_id |
| POST | /api/v1/vantage/query |
| GET | /api/v1/knox-evidence |
| GET | /api/v1/knox-brief |
Embed Tools
Drop a live KC badge onto any website or CMS with a single script tag. No API key required.
View embed docs →Customize widget colors, typography, and border radius. Copy a pre-configured script tag with your brand settings encoded inline.
Open Theme Builder →More developer resources
StatVault runs a live Model Context Protocol server. Point any MCP host at POST /api/mcp and call statvault_grounded_rating or statvault_career_standing. Every result carries provenance, and a tool that cannot verify an answer returns isError:true with a reason instead of guessing — which is the whole point of citing us. Tool shapes for OpenAI function-calling live at /api/v1/tools.