A3S Lab

Tutorials

Step-by-step guides for building production AI agents with A3S.

Build a Multi-Agent Review Flow

Use A3S Code task delegation, .a3s/agents, automatic subagents, and run inspection from the Node SDK.

Build a Contract Review Agent

Use A3S Code structured output to build an autonomous contract review agent that produces deterministic JSON reports.

Build a Research Agent

Compose A3S Code v3.1.0 programmatic tool calling, task delegation, streaming, and queue inspection with the Node SDK.

Hooks

Run custom logic before and after tool executions with the hook system.

MCP Integration

Connect external tool servers to your agent using the Model Context Protocol.

Memory System

Give your agent long-term memory that persists across sessions.

Build a News Radar Workflow

Use A3S Code v3.1.0 MCP stdio registration, direct tools, and streaming with clear live-source boundaries.

Programmable Orchestration

Express deterministic multi-agent orchestration in code with parallel fan-out, pipelines, and resumable workflows.

Programmatic Tool Calling (PTC)

Replace multi-turn LLM tool loops with bounded, deterministic JavaScript programs using session.program().

Security & Human-in-the-Loop

Control agent permissions with policies, confirmation flows, and HITL patterns.

Session Persistence

Save, resume, and manage long-running agent sessions across restarts.

Skills System

Extend agent capabilities with reusable, composable skills — prompt injections, personas, and tool-gated behaviors.

Streaming & Real-time

Build real-time applications with session.stream() — token-by-token output, tool progress, and event-driven UIs.

Structured Output

Use generate_object to produce deterministic, schema-validated JSON from any LLM provider.