[FEATURE]: OpenCode with Prime Agent style thinking

Feature hasn't been suggested before.

• I have verified this feature I'm about to request hasn't been suggested before.

Describe the enhancement you want to request

I came across Prime Agent and the underlying Recursive Language Model (RLM) framework, and I’m wondering whether OpenCode could integrate some of these ideas — even if not the full framework. It is saturating the ARC AGI 3 - https://x.com/PrimeIntellect/status/2085087000764568010
Prime Agent’s approach is interesting because its a self‑improving coding harness built on RLM principles:
persistent REPL

• recursive LM calls for context decomposition
• context treated as a manipulable variable
• sub‑agents as first‑class function calls
• automatic compaction with harness rewrites

Full write‑up: https://www.primeintellect.ai/blog/prime-agent
Original RLM explanation by Alex Zhang: https://alexzhang13.github.io/blog/2025/rlm

Is it feasible for OpenCode to adopt RLM‑style thinking — persistent REPL, recursive calls, context compaction, sub‑agent spawning — without fully switching to the Prime Agent framework?