KyroDB docs
Build with fresh, provable context.
Install the SDKs, call the runtime API, wire change events, and operate evidence without exposing tokens or serving stale context silently.
Connect pgvector, retrieve a fresh scoped ContextPacket, send change events, and inspect evidence.
Read guideRuntime credentialsFind, copy, and store the server-side runtime environment values safely.
Read guidePython SDKInstall `kyrodb==1.0.2` from trusted server-side Python code.
Read guideTypeScript SDKInstall `kyrodb@1.0.2` from Node/server-side TypeScript code.
Read guideRuntime APIHTTP routes, auth planes, request bodies, and response contracts.
Read guideFreshness and scopeHow KyroDB decides whether context is fresh, scoped, and safe to serve.
Read guideConnectorspgvector, Qdrant, vector database, and custom-store connector requirements.
Read guideProof and replayTrace diagnosis, feedback, proof bundles, replay, and shadow adoption for context reliability.
Read guideFirst runtime call
HTTPcurl -X POST "$KYRODB_BASE_URL/v1/context/retrieve" \
-H "Authorization: Bearer $KYRODB_DATA_PLANE_TOKEN" \
-H "Content-Type: application/json" \
--data '{
"query_embedding": [0.92, 0.08, 0.02, 0.01],
"scope": { "tenant_id": "acme", "namespace": "kb" },
"top_k": 3,
"freshness_mode": "strict"
}'