{"name":"ckg-nvidia-ai","description":"Auditable knowledge graph for the full NVIDIA AI developer stack across 20 domains. Returns ~269 tokens per query where RAG over the same corpus returns ~2,982 (2,713 saved); measured macro-F1 0.471 vs 0.123. $0.010 USDC per call via x402 — break-even at $3.69/MTok input price. Every edge is authored, source-linked and SHA-256 verifiable.","url":"https://ckg-nvidia-ai.onrender.com","version":"0.23.0","protocolVersion":"1.0","authentication":{"schemes":["none"]},"capabilities":{"streaming":false,"pushNotifications":false,"stateTransitionHistory":false},"skills":[{"id":"list_domains","name":"list_domains","description":"List all 20 NVIDIA AI domains available in this CKG."},{"id":"search_concepts","name":"search_concepts","description":"Find concepts in a NVIDIA AI domain by keyword."},{"id":"query_ckg","name":"query_ckg","description":"Traverse the NVIDIA AI knowledge graph from a concept outward."},{"id":"get_prerequisites","name":"get_prerequisites","description":"Full upstream prerequisite chain for a NVIDIA AI concept."},{"id":"ask_nvidia","name":"ask_nvidia","description":"Natural-language query over the NVIDIA AI stack."},{"id":"list_ecosystem","name":"list_ecosystem","description":"List all nodes in a NVIDIA AI domain with taxonomy and source URLs."}],"economics":{"price_usd_per_call":0.01,"mean_tokens_returned":269,"baseline_mean_tokens":2982,"baseline_method":"retrieval-augmented generation over the same corpus","tokens_saved_per_call":2713,"breakeven_input_price_usd_per_mtok":3.69,"answer_quality_macro_f1":0.471,"baseline_macro_f1":0.123,"decision_rule":"This call costs $0.010 USDC and returns ~269 tokens instead of the ~2,982 a RAG retrieval over the same corpus returns, saving ~2,713 input tokens. It pays for itself on token cost alone when your input price exceeds $3.69 per million tokens. Below that, invoke only when answer quality matters: measured macro-F1 is 0.471 versus 0.123 for RAG on identical queries.","traceability":"Every returned edge is authored, not inferred, and carries a source URL; where source provenance is present it also carries a SHA-256 of the source bytes, so any answer can be checked against the document it came from.","benchmark":{"version":"v0.6.2","domains":44,"queries":7758,"url":"https://github.com/Yarmoluk/ckg-benchmark","dataset":"https://huggingface.co/datasets/danyarm/ckg-benchmark","reproducible":true}},"payments":{"protocol":"x402","amount_usd":"0.010","token":"USDC","chain":"base","recipient":"0x9B987263C9Da951E9044D58f93f1940c5dF1cF1B","extensions_uri":"https://x402.org/extensions/a2a/v1"},"publisher":"Graphify.md","publisher_url":"https://graphify.md"}