Back to directory

OpenCode Orchestrator

An OpenCode plugin that keeps a task and its unfinished work connected across turns, using a Commander role for the overall goal and optional Planner, Worker and Reviewer roles when a task is complex enough to benefit.

OpenCode Orchestrator

OpenCode Orchestrator adds persisted, resumable missions to OpenCode. Starting a mission with a command such as /task keeps the goal and its state in the project's .opencode directory across turns, replacing any active mission in the same project, while existing project TODO and checklist files stay available for the new mission to reconcile.

Ordinary questions and small edits are handled directly by Commander without invoking the other roles, which are used only when they help rather than as a required pipeline; a /plan command prepares a plan without implementing it, and /stop or /cancel ends the active mission. Configuration options, such as how many worker agents can run concurrently, are set in OpenCode's own configuration file.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.