Skip to content

CLI Reference

The rax CLI provides project scaffolding, agent generation, and agent execution.

Create a new Reactive Agents project.

Terminal window
rax init <name> [--template minimal|standard|full]

Templates:

TemplatePackages Included
minimalcore, runtime, llm-provider
standard+ memory, reasoning, tools
full+ guardrails, verification, cost, orchestration, prompts

Generate an agent file from a recipe.

Terminal window
rax create agent <name> [--recipe basic|researcher|coder|orchestrator]

Recipes:

RecipeWhat It Generates
basicMinimal agent with LLM only
researcherAgent with memory + reasoning
coderAgent optimized for code tasks
orchestratorMulti-agent orchestrator with memory

Run an agent with a prompt.

Terminal window
rax run <prompt> [--provider anthropic|openai|ollama|test] [--model <model>] [--name <name>]

Example:

Terminal window
rax run "Explain quantum computing" --provider anthropic --model claude-sonnet-4-20250514

Start an agent as an A2A server.

Terminal window
rax serve [--port <number>] [--name <name>] [--provider <provider>] [--model <model>]
[--with-tools] [--with-reasoning] [--with-memory]

Options:

OptionDefaultDescription
--port3000HTTP port for the A2A server
--name"agent"Agent name (used in Agent Card)
--provider"test"LLM provider
--modelModel name
--with-toolsoffEnable built-in tools on the A2A server agent (file-write, web-search, etc.)
--with-reasoningoffEnable reasoning strategies
--with-memoryoffEnable memory (tier 2)

Endpoints served:

  • GET /.well-known/agent.json — Agent Card (A2A discovery)
  • GET /agent/card — Agent Card (fallback)
  • POST / — JSON-RPC 2.0 (message/send, tasks/get, tasks/cancel, agent/card)

Example:

Terminal window
rax serve --name researcher --provider anthropic --model claude-sonnet-4-20250514 --with-tools --port 4000

Fetch and display the Agent Card from a remote A2A-compatible agent server.

Terminal window
rax discover <url>

Fetches GET <url>/.well-known/agent.json and pretty-prints the agent’s name, description, capabilities, and supported skills.

Example:

Terminal window
rax discover http://localhost:3000
Agent Card: researcher
Provider: anthropic (claude-sonnet-4-20250514)
Capabilities: streaming, tools
Skills: web-search, file-write
Endpoint: http://localhost:3000

Start the development server (placeholder).

Terminal window
rax dev

Run an evaluation suite (placeholder).

Terminal window
rax eval run --suite <suite-name>

Launch interactive REPL (placeholder).

Terminal window
rax playground

Inspect agent state (placeholder).

Terminal window
rax inspect <agent-id> [--trace last]
Terminal window
rax version
rax --version
rax -v
Terminal window
rax help
rax --help
rax -h