Skip to content

MAS Framework — Hands-On Tutorials

Progressive tutorials for the MAS Framework — from single-agent basics to full lab experiments. They are one of several ways to work with MAS Lab; you can also reproduce paper labs or use the Web UI to design and inspect agents and experiments.

Start with Tutorial 0 — Docker or developer install, LLM credentials, and verification. Same setup for CLI and web UI.

# Tutorial What you learn
0 Environment Setup Install, PATH, model endpoint, API key wiring
1 Build an agent Agent manifest, tools, skills, memory, CLI, traces
2 Orchestrate your MAS MAS manifest, delegation, topology overlays
3 Run an experiment Experiments, benchmarks, pipelines, MCEv1 evaluation
4 MCP tools Discover and dispatch local and remote tools through a stable contract
5 A2A agents Discover and dispatch local and remote agents through a stable contract
6 Agent skills Teach an agent a reusable skill — procedure, reference material, checked script — without folding it into the agent's own instructions
8 Telemetry Native traces, live OTel, replay, collector / ClickHouse
9 KG & OXP Native→KG, OTel→KG (OXP norm), Neo4j serialization
10 Evaluation metrics Stock MCE, one-off prompts, reusable EvalMetrics, judge infra
11 Sessions and recovery Checkpoint, fork, resume and recover a conversation
12 Spawned subagents Run bounded, pre-authored subagent templates during a turn
13 Control attach and debug scripts Interrupt a bad session, persist, resume; investigate; steer-to-fix on the restarted chat; auto-checkpoint tree; gdb script_file / --debug-script
14 Governance and HITL Require human or automated-policy approval before a tool call runs

After Tutorial 14, reproduce all paper results across 3 labs: Paper.

Start from Tutorial 0 for install and first commands — CLI and optional web UI use the same setup.

Quick start

See Tutorial 0 — Environment setup for Docker and developer install paths, LLM credentials, and verification steps.

Tutorials 0–3 ship demo/scenario.yaml files with structured steps, commands, and expected output used by tests/tutorials/test_scenario_commands.py.

Replay all offline commands with logs (stdout/stderr per tutorial under /tmp):

python scripts/run_tutorial_scenarios.py
# Live LLM steps too:
TUTORIAL_ONLINE=1 python scripts/run_tutorial_scenarios.py
# Logs: /tmp/tutorial-00.log … /tmp/tutorial-03.log

Telemetry is Tutorial 8 (mas-library-telemetry). Knowledge-graph normalization is Tutorial 9 (mas-library-kg): native→KG and OTel→KG (OXP norm wrap).

What's next

After Tutorial 14: