REST API
Versioned tenant-scoped responses, sources, evals, releases, actions, structured errors, and hard usage caps.
Developers
One protocol layer for humans and agents.
Build with REST, TypeScript, Python, webhooks, declarative manifests, CLI, and scoped MCP tools.
A stable API with evidence built in.
Every successful answer returns current evidence, retrieval confidence, a durable conversation ID, and token usage.
- OpenAPI 3.1 contract
- Hashed scoped credentials
- Idempotent action receipts
- Explicit abstention
{ "input": "How do I enforce SSO?"}{ "data": { "status": "answered", "confidence": 0.91, "evidence": [{ "title": "Security guide" }] }}Management MCP
Create one revocable client token per app, choose its scopes, and keep every call isolated to its workspace.
Verified AI MCP
Let AI clients inspect sources and living docs, track editorial work, create governed drafts, and ask evidence-grounded questions through Streamable HTTP.
kiteframe_sources
kiteframe_docs
kiteframe_publish_articles
kiteframe_create_article
kiteframe_answer
TypeScript and Python
The official @sansavision/kiteframe package exposes answers, sources, Living Docs, Fieldnotes, domains, and delivery with scoped bearer credentials.
CLI
Ask verified knowledge and inspect sources from local or CI environments.
Webhooks
Encrypted signing secrets, HMAC-SHA256 signatures, delivery health, and replay-safe event identifiers.
Official TypeScript SDK
One client for every Kiteframe surface.
Install the Sansa Vision package, create a scoped API key, and use the same typed client for verified answers, docs, publishing, domains, and exports.
Create a sandboxnpm install @sansavision/kiteframeimport { Kiteframe } from "@sansavision/kiteframe";const kiteframe = new Kiteframe({ apiKey: process.env.KITEFRAME_API_KEY});const answer = await kiteframe.ask({ question: "Which plan fits a beginner?", sources: "required"});