AI Coding Dojo
Watch the coding agent's plan, diffs, and blast radius live.
The blast-radius heat-map turns 'the agent edited my code' from a vague worry into a literal, file-by-file diff map — including the files it read but chose not to touch.
What goes in, what comes out
Learner gives a coding agent a scoped spec against a real sandboxed repo; the agent's plan, file reads, and proposed diffs stream into a glass-box trace panel instead of silently landing in the repo. Learner approves, edits, or rejects each diff individually, and a repo heat-map colors every file by touched, considered-but-not-touched, and never-seen so the actual blast radius is visible before merge.
A written spec or ticket-style task description, and a starter repo (seeded template or learner's own scoped upload).
A step-by-step agent trace, an approve/reject diff queue, a blast-radius heat-map of the repo, and a final pass/fail against the spec's own test suite.
Coding agent/model, autonomy level (propose-only vs. auto-apply-with-undo), spec strictness, repo size/seed template, and a step budget.
In-module lab for AIE-201 (MUST tier); doubles as the F5 ship-one-checked-output artifact in Core Foundation.
Go deeper, elsewhere
Hand-picked public explainers and open tools that complement this one — always optional, never required, never graded.
Concepts click when you open the machinery.
Three labs are already live and free — the same hands-on style this playground brings to its module.