Twilio AI Agent Architect for Claude Code: Plan a Support Assistant
Use Twilio's planning skill to map entry channels, allowed answers and transfer criteria for a conversational support agent before choosing an integration.
中文Setup for Claude Code
Claude Code and a text brief in a scratch workspace. The example uses fictional conversations and does not require a Twilio account.
- In Claude Code add the Twilio marketplace with `/plugin marketplace add twilio/ai` and install `/plugin install twilio-developer-kit@twilio`. Invoke /twilio-ai-agent-architect.
- Write down the channel, the agent's scope, and which outcomes must hand off to a person before choosing a Twilio integration path.
Example task
Describe a fictional library assistant that answers opening-hours questions on the phone. Feed two mock requests: 'What time do you open?' and 'Change my account details.' Ask for the planned outcome of each and the boundary between chat logic and human support.Expected result
The first request can use a supplied opening-hours answer. The account update routes to staff because it requires a separate trusted workflow. The plan accounts for both outcomes explicitly.
Upstream documentation
Original SKILL.md on GitHub · MIT license · Source reviewed
Frequently asked question
What does the Twilio AI Agent Architect skill decide?
It maps your users, channels and outcomes to a Twilio conversational architecture and identifies the implementation areas you need to verify.