io.github.yasmanycastillo/model-shunt
model-shunt-mcp
Model routing for AI agents: delegate bulk reads & boilerplate to cheap worker models.
alive probed 2026-10-15
3tools
762tokens in Claude Opus 5.5
792tokens in Claude Code
0.4 %of a 200k context window
451o200k_base estimate
321 msto initialize
2 msto list tools
- Run it
- uvx model-shunt==1.2.1 (stdio, probed in a container with the network off)
- Host
- pypi.org
- Reports itself as
- model-shunt-mcp 1.2.1
- Capabilities
- tools
- Instructions
- none
- Claude count
- Claude Opus 5.5, measured 2026-10-15. Counted again when the tool definitions change. Method.
- Repository
- https://github.com/yasmanycastillo/model-shunt
- Registry
- active, updated 2026-09-24 · registry entry
- Reachable
- 1 of the last 1 probe
Most expensive tools
| Tool | Tokens (o200k) | Share of this server |
|---|---|---|
| code_write | 183 | 41 % |
| bulk_read | 170 | 38 % |
| get_available_models | 98 | 22 % |
Badge and feed
[](https://mcpcost.ulehla.dev/s/io.github.yasmanycastillo/model-shunt)
An entry for every probe that changed the result, the tool count or the token cost: Atom feed. Add it to a feed reader or a Slack RSS app to hear when the server stops answering or its tool set grows.