Prunes historical reasoning parts from the message history sent to the LLM (keeping the last turn's), saving context window and tokens on reasoning models. Model-agnostic; uses the idiomatic experimen
Version 1.0.0, 8 weekly downloads, rank 1400. Judged by scanning its code (heuristic).
| @opencode-ai/plugin | - |
|---|---|
| @opencode/plugin | - |
[](https://rafal-one.github.io/opencode-v2-ready/p/ghilteras__opencode-reasoning-pruner.html)
It depends on the V1 plugin API and will not load in OpenCode 2 until the author ports it. See how to port.