OpsfleetAgent Bootcamp

Finish

Clean up

Remove everything you provisioned, and confirm nothing is left running on your name.

10 minEasy
Alice’s ask

The “Leave No Trace” finish

Alice's assistant is live, observable and governed. Before you leave, tear your stack down: idle runtimes, gateways and memories still exist until you remove them.

You

What you do as a developer

  1. Tear down your stack

    terminal
    uv run bootcamp.py down

    Usually takes ~10.5 min. It destroys every stage you provisioned (agent, evaluators, Gateway with its weather Lambda and Cedar policy, Memory, MCP runtime, Cognito and the story users, your code bucket and its database copy, and the optional guardrail), then deletes what terraform does not track: your golden dataset and batch evaluations (created by eval) and your runtime and evaluation-result log groups.

  2. Verify

    terminal
    uv run bootcamp.py status
    expected outputread only
    No stack provisioned for '<you>' (start with `uv run bootcamp.py up 0`).

  3. What stays

    Your participant role bootcamp-participant-<you>, your LiteLLM key and budget, and the secret bootcamp/participants/<you>/litellm belong to the instructor's account stack: you can't delete them and don't need to. Your local files (phase1/datastream_corp.db, phase1/sessions/, .env) stay on your laptop.

Check your work

status prints No stack provisioned. If down stopped with an error, run it again: it picks up where it left off. If your instructor removed your stack for you (instructor.py reset), run uv run bootcamp.py forget instead to drop your stale local terraform state.

Under the hood

down runs terraform destroy on your participant workspace, then deletes the log groups AgentCore created for your runtimes and evaluations (/aws/bedrock-agentcore/runtimes/awsworkshop_<you>_* and /aws/bedrock-agentcore/evaluations/results/awsworkshop_<you>_*). After the workshop the instructor checks for leftovers with instructor.py purge before removing the account stack.