🚀 Zero Polling • Free Models • AI-Powered

Code Smarter,
Not Harder

Run OpenCode AI coding agent on OpenClaw without token waste. Dispatch tasks, get results, build faster.

Get Started → Learn More

opencode-dispatch

$ ~/.openclaw/scripts/dispatch-opencode.sh \

-p "implement auth API" \

-n "auth-api" \

-w "/project" \

-m "opencode/glm-5-free"

🚀 Starting: auth-api

✅ auth-api → done

📄 /root/.openclaw/data/opencode-results/auth-api.json

Why OpenCode on OpenClaw?

Four powerful methods for different scenarios

Zero Polling

Dispatch tasks without wasting tokens on status checks. Results saved to JSON.

Free Models

Use opencode/glm-5-free or minimax-m2.5-free. No API costs.

Multi-Task DAG

Combine with ralph-team for complex projects with parallel execution.

Background Run

Tasks run in background. Continue chatting while code gets written.

Quick Start

Three ways to use OpenCode on OpenClaw

METHOD 1

Dispatch (Recommended)

                        
# Zero polling - save result to JSON
~/.openclaw/scripts/dispatch-opencode.sh \
  -p "implement auth API" \
  -n "auth-api" \
  -w "/project"
                        
                    

METHOD 2

Interactive PTY

                        
# With pseudo-terminal
bash pty:true workdir:/project \
  command:"opencode run 'task'"

METHOD 3

Team Tasks DAG

                        
# Multi-task with dependencies
TM="team-tasks"
$TM init project -m dag
$TM add project "task1"
$TM add project "task2" -d "task1"
                        
                    

Available Models

GLM-5-Free

opencode/glm-5-free

✅ Recommended

MiniMax M2.5

opencode/minimax-m2.5-free

Free

Gemini Flash

google/gemini-1.5-flash

Free

Ready to Code Smarter?

Install OpenCode and start building faster with AI assistance.

View on GitHub