Opencode Controller

This is a workflow enforcer for Opencode that keeps Claude from writing code directly and instead orchestrates it through Opencode's plan and build agents. It handles the session setup, model selection, and authentication flow, then enforces a strict loop: always plan first, get approval, then build. If the build agent hits a question, it forces you back to planning. The core insight here is useful: separating planning from execution prevents half-baked implementations. You'll want this if you're already using Opencode and find yourself or Claude skipping the planning step or trying to answer design questions mid-build. It's basically guard rails for a two-phase workflow.

skills.sh installs: 166
GitHub stars: 674
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add sundial-org/awesome-openclaw-skills --skill opencode-controller --agent claude-code

Installs into .claude/skills of the current project.

Files

Opencode Controller

Core rule

Clawdbot does not write code. All planning and coding happens inside Opencode.

Pre-flight

Ask the user which AI provider to use.
Ask how the provider should be authenticated.
Do not proceed without confirmation.

Session management

Start Opencode.
Open session selector using: /sessions
If the current project already exists:
- Select the existing session.
Never create a new session without user approval.

Agent (mode) control

Open agent selector using: /agents
Available agents:
- Plan
- Build
Always select Plan first.
Switch agents whenever required using /agents.

Model selection

Open model selector using: /models
Select the user-requested provider.
If authentication is required:
- Copy the login link provided by Opencode.
- Send it to the user.
- Wait for confirmation before continuing.