opencode-cli

name	opencode-cli
description	Reference for using OpenCode CLI to handle complex coding tasks. Use when Research-Claw needs to delegate multi-file projects, code generation, or iterative debugging to a multi-provider AI coding tool. Trigger words: opencode, coding CLI, delegate coding, multi-provider agent.

OpenCode CLI

Terminal-based AI coding agent supporting multiple LLM providers (OpenAI, Anthropic, Google, local models). Features a TUI for interactive use and run command for automation.

Detect Installation

which opencode  # should return a path
opencode --version

Install

# Recommended (automated)
curl -fsSL https://opencode.ai/install | bash

# npm
npm install -g opencode-ai

# Homebrew (macOS/Linux)
brew install anomalyco/tap/opencode

# pnpm
pnpm install -g opencode-ai

Requires API keys for your chosen LLM provider (configured via /init on first run).

Non-Interactive Usage (run)

For programmatic invocation from Research-Claw, use opencode run:

# Basic non-interactive execution
opencode run "Refactor the auth module to use dependency injection"

# With model selection (provider/model format)
opencode run -m anthropic/claude-sonnet-4-20250514 "Fix all type errors"

# Attach files for context
opencode run -f src/config.ts -f README.md "Update the config schema"

# JSON output for parsing
opencode run --format json "List all exported functions"

# Continue a previous session
opencode run -c "Now add tests for what you just implemented"

# Continue a specific session
opencode run -s <session-id> "Fix the remaining issues"

Key Flags

Flag	Description
-m, --model <provider/model>	Model in provider/model format
-f, --file <path>	Attach file(s) to the message (repeatable)
--format <fmt>	Output format: default or json
-c, --continue	Continue last session
-s, --session <id>	Continue a specific session by ID
--fork	Fork session when continuing (new branch)
--share	Share the session
--agent <name>	Specify which agent to use
--attach <url>	Connect to running server (e.g. http://localhost:4096)
--log-level <level>	DEBUG / INFO / WARN / ERROR

Research-Claw exec Patterns

# Delegate a complex coding task
exec("opencode run -m anthropic/claude-sonnet-4-20250514 'Implement the feature described in TODO.md'")

# With file attachments for context
exec("opencode run -f README.md -f src/schema.ts 'Add validation based on the schema'")

# JSON output for structured results
exec("opencode run --format json 'List all API endpoints in this project'")

# Using persistent server for multiple tasks
exec("opencode run --attach http://localhost:4096 'Run the test suite and fix failures'")

Docs

• CLI docs: https://opencode.ai/docs/cli/
• Getting started: https://opencode.ai/docs