According to official OpenCode docs, they provide an action we could use with Kimi K2 to run the agentic loop like we do with Claude Code and OpenAI Codex.
I believe that should greatly improve the reviews produced as it would now be able to run tests, linters, provers, etc, on demand, use sub-agents to handle big volume PRs and access project-specific plugins if we were to configure them.
According to official OpenCode docs, they provide an action we could use with Kimi K2 to run the agentic loop like we do with Claude Code and OpenAI Codex.
I believe that should greatly improve the reviews produced as it would now be able to run tests, linters, provers, etc, on demand, use sub-agents to handle big volume PRs and access project-specific plugins if we were to configure them.