io.github.Yarmoluk/ckg-nvidia-ai
NVIDIA AI CKG
io.github.Yarmoluk/ckg-nvidia-ai costs 1 605 tokens of context per session. It has 9 tool definitions. That is 0.8 % of a 200k window. Measured 2026-10-03.
NVIDIA AI knowledge graphs — 20 domains. 4x F1, 11x fewer tokens, SHA-256 provenance. MCP-native.
alive probed 2026-10-03 · HTTP 200
9tools
1.6ktokens for the tool definitions
0.8 %of a 200k context window
1.4 sto initialize
90 msto list tools
- Endpoint
- https://ckg-nvidia-ai.onrender.com/mcp (streamable-http)
- Host
- ckg-nvidia-ai.onrender.com
- Protocol
- 2025-11-25
- Reports itself as
- ckg-nvidia-ai 1.30.0
- Capabilities
- experimental, prompts, resources, tools
- Instructions
- 1 012 characters
- Repository
- https://github.com/Yarmoluk/ckg-nvidia-ai
- Registry
- active, updated 2026-07-29 · registry entry
- Reachable
- 2 of the last 2 probes
Most expensive tools
| Tool | Tokens | Share of this server |
|---|---|---|
| query_intersect | 483 | 30 % |
| route_query | 271 | 17 % |
| ask_nvidia | 188 | 12 % |
| query_ckg | 170 | 11 % |
| verify_source | 162 | 10 % |
History
We add a row when the result changes. A server with no change keeps its last row.
| Probed | Result | Init | Tools | Tokens |
|---|---|---|---|---|
| 2026-10-03 20:36 | alive | 1.4 s | 9 | 1.6k |
| 2026-10-03 17:52 | alive | 608 ms | 9 | 1.6k |
Badge
[](https://mcp-pulse.ulehla.dev/s/io.github.Yarmoluk/ckg-nvidia-ai)