Supported Agents
Blueprint AI supports multiple AI agents. You can switch between them per session or run multiple agents in parallel across tabs. Selecting a different runtime from the agent picker opens a new tab automatically. Runtime and ACP installs are manual; Blueprint AI does not auto-download dependencies. Select an agent below for setup instructions.
- Selecting a different agent from the runtime picker opens a new tab for that runtime.
- Your current chat tab is preserved, so you can keep multiple concurrent sessions (for example, one Claude tab and one Codex tab).
- In the agent picker, each runtime shows a status label such as Installed, Not installed, or Needs API key.
# Available Agents
Built-in HTTP agent with 200+ models. No CLI or external process required.
Get API key from openrouter.ai, paste in Project SettingsAnthropic's coding agent via JSON-RPC over stdio. Supports session resume and thinking.
curl -fsSL https://claude.ai/install.sh | bash && npm install -g @anthropic-ai/claude-code-acpGoogle's Gemini agent with built-in ACP support. Requires Google Cloud auth.
npm install -g @google/gemini-cliOpenAI's Codex agent with reasoning effort configuration.
npm install -g @openai/codex && npm install -g @anthropic-ai/codex-acpGitHub Copilot with built-in ACP support. Requires GitHub auth.
npm install -g @github/copilotGo-based AI coding agent with self-contained ACP support.
npm i -g opencode-ai# Custom Agents
You can add custom ACP-compatible agents via Project Settings > Plugins > Blueprint AI > BAI Agents > Custom Agents.
Configuration includes display name, executable path, arguments, working directory, environment variables, and optional API key or model ID.
# CLI Agent Requirements
- Node.js 18+ — Required for all CLI agents. The plugin auto-detects Node.js from common install paths and version managers (nvm, volta, fnm, bun, mise).
- CLI agents run as external processes and communicate via JSON-RPC over stdio.
- Blueprint AI does not auto-download CLIs or ACP adapters. Install required runtimes manually in your terminal.
# Status Labels
The runtime picker shows a live status for each agent: Installed, Not installed, or Needs API key. Use this as a quick health check before sending a prompt.