Skip to content

Tutorials#

Three walk-throughs. Each is a Jupyter notebook rendered inline; all run end-to-end against a local swarm instance.

Unlike the how-to guides which solve one problem at a time, tutorials are contextual storytelling — a realistic scenario from beginning to end.

  • End-to-end fraud classifier

    BFSI scenario from scratch. Synthetic transaction data → pipeline run with rbi_free_ai compliance → shadow traffic → promotion → audit PDF. ~30 minutes.

  • Drift investigation

    A production model's nightly drift check fires. Walk through triage, root-cause, and remediation. ~20 minutes.

  • Plugin authoring

    Build a hello-world CC-format plugin with a skill + command + agent + hook. Install, verify, uninstall. ~20 minutes.

Running the notebooks#

Just click. MkDocs renders them inline — all outputs visible, code copyable.

git clone https://github.com/TheAiSingularity/swarm.git
cd swarm
.venv/bin/pip install jupyter
.venv/bin/jupyter notebook docs/content/tutorials/

Notebooks expect a running swarm stack (see Quickstart) and an ANTHROPIC_API_KEY or OPENAI_API_KEY in the environment.

Each notebook has an "Open in Colab" button (rendered only on the published site, not in the raw .ipynb).