@loopstack/code-agent-example-workflow

Demonstrates how to launch the ExploreAgentWorkflow from @loopstack/code-agent as a sub-workflow to explore a remote workspace and surface a synthesized summary.

By using this example you’ll get…

A parent workflow that runs ExploreAgentWorkflow with a fixed exploration brief
A LinkDocument that renders the embedded sub-workflow while it runs
A final MessageDocument summarizing the agent’s findings

Installation


npm install @loopstack/code-agent-example-workflow

@loopstack/code-agent transitively requires @loopstack/claude-module and @loopstack/remote-client — make sure both are configured (Anthropic API key + sandbox environment) before running.

How It Works

The workflow starts and calls ExploreAgentWorkflow.run({ instructions }).
The sub-agent iterates over glob, grep, and read tool calls against the remote workspace until it reaches a final answer.
The callback fires with the agent’s synthesized response.
A MessageDocument is saved with the response text.

Public API

CodeAgentExampleModule
CodeAgentExampleWorkflow

Dependencies

@loopstack/common
@loopstack/code-agent