Index

Labs

Educate by doing. Lab 01 attacks a bare agent, repeats the attack inside an Agent Substrate actor, and scores the difference from the cluster. All lab attacks run only on a local kind cluster.

Lab 01Draft2 hoursintermediate
Secret Exfiltration
A local kind lab that shows why an AI agent should not hold a long-lived API key, and how Agent Substrate keeps that key out of the actor.
Lab 02Coming soon
Egress Allowlist
Placeholder. Learn which outbound calls an allowlist permits, and which ones stop at the boundary. Runnable steps are coming in the lab.
Lab 03Coming soon
Blast Radius
Placeholder. Compare how far a steered agent can reach with no boundary, and what a sandbox leaves it able to do. Runnable steps are coming in the lab.

Add a lab

A published lab is a folder under labs/. The page shape is thetemplate. Lab 01 is the example.

Add labs/NN-short-name/lab.md with title, slug, status, and optional duration and difficulty. The site shows that status.draft is not "coming soon".
Add labs/NN-short-name/README.md. A single-page lab keeps these level-2 headings, in order: Objectives, Prerequisites, Setup, Attack, Defend, Measure, Cleanup. A walkthrough lab keeps the README as a short overview and adds steps/01-name.md with Next and Back links. The site renders the overview at /labs/NN-short-name/ and each step beside it. Lab 01 is the walkthrough.
Remove any placeholder at site/src/content/labs/NN-short-name.md. A placeholder is only for a lab that does not have a repo folder yet.
From site/, run npm run build. The build fails if a required heading is missing, or if site copy contains an em dash or an en dash.

Labs

Secret Exfiltration

Egress Allowlist

Blast Radius

Add a lab