AI AGENTS
ProtoAgent
by Thomas Gauvin
Overview
Minimal TypeScript AI coding agent CLI with tool-use loops, approvals, sessions, MCP, skills, sub-agents, and a build-your-own tutorial.
Details
ProtoAgent is a minimal TypeScript AI coding agent CLI by Thomas Gauvin. The supplied website, documentation, and GitHub metadata describe it as supporting tool-use loops, approvals, sessions, MCP, skills, sub-agents, and a build-your-own tutorial. It is relevant for developers evaluating how a compact coding-agent CLI is structured or looking for a TypeScript-based agent project to inspect.
When to Use
Use ProtoAgent when you want to evaluate a minimal TypeScript CLI for AI coding-agent workflows. Use it to study agent patterns such as tool-use loops approval flows sessions MCP integration skills and sub-agents. Use the reference specification and GitHub repository when comparing small coding-agent projects or learning how to build your own agent.
Getting Started
- Open the official website at https://protoagent.dev/ for the project overview.
- Read the reference specification at https://protoagent.dev/reference/spec.
- Inspect the GitHub repository at https://github.com/thomasgauvin/protoagent.
- Review the build-your-own tutorial mentioned in the supplied project metadata before adapting ideas into another project.
Key Features
- •Minimal TypeScript AI coding agent CLI.
- •Tool-use loops for coding-agent workflows.
- •Approval support for agent actions.
- •Session support.
- •MCP support.
- •Skills and sub-agents.
- •Build-your-own tutorial.
Capabilities
- •AI coding agent CLI
- •tool-use workflows
- •approval workflows
- •session-based agent workflows
- •MCP integration
- •skills
- •sub-agents
Last updated Jun 27, 2026