New Live Workshop: Hands-On Loop Engineering with Claude
Most people use Claude through individual sessions: provide some context, ask it to complete a task, review the result, and repeat the same process the next time.
This works well for isolated tasks, but it becomes inefficient when the workflow is repetitive, long-running, or requires Claude to keep track of progress across multiple runs.
Loop engineering addresses this by giving the workflow a clearer structure. Instead of repeatedly issuing disconnected prompts, you define how Claude should inspect the current state, choose the next action, execute it, verify the result, record progress, and stop when the task is complete or requires human review.
I am hosting a new 3-hour live workshop on Hands-On Loop Engineering with Claude, where we will cover both the design principles and the practical implementation of these workflows.
During the workshop, we will cover:
What a Claude loop actually is
When you should—and should not—build a loop
The minimum viable Claude loop
Designing clear goals, boundaries, and stopping conditions
Adding persistent state with PROGRESS.md
Tracking completed work, failures, and the next action
Separating the worker from the checker
Adding tests, validation, and review steps
Scheduling repeated workflows with /loop
Adding tools, connectors, and safer permissions
Preventing runaway loops, repeated failures, and unnecessary token usage
Building a complete Claude loop from scratch
The workshop is divided into three parts:
0:00–0:45 — Theory
We will begin with the core ideas behind loop engineering, including state design, verification strategies, permissions, stopping conditions, and the situations where loops are useful or risky.
0:45–1:45 — Guided practical
We will build a working Claude loop step by step, starting with a simple repeated instruction and gradually adding persistent state, verification, progress tracking, and scheduling.
1:45–2:00 — Break
2:00–3:00 — Attendee lab
You will begin designing and building your own loop based on a workflow that matters to you. This could be a coding, documentation, issue-triage, research, content, or automation workflow.
During this section, I will review attendee implementations, answer questions, and help refine the loop’s scope, state structure, verification process, and safety boundaries.
Who should attend?
This workshop is suitable for developers, AI engineers, researchers, technical writers, and builders who already use Claude or Claude Code and want to move beyond one-off prompting.
You do not need advanced experience with agent frameworks. However, you should be comfortable working with prompts, files, terminals, and basic developer workflows.
What you will receive
Attendees will receive:
Access to the 3-hour live workshop
A reusable Claude loop template
State and progress file examples
Verification and safety checklists
Practical loop examples
Access to the workshop recording
Supporting materials and resources
Workshop Details
Workshop date: Tuesday 21 st July
Time: 19:00 - 22:00
Standard ticket: 50$