Hi,
I am running two live workshops this month, both on Zoom, both recorded, both with slides.
A Skill packages a procedure Claude Code can load when it becomes relevant, instead of one you re-explain at the start of every session. The instructions, the structure, and the description that decides when it triggers all live in files you version alongside the project.
Skills usually fail for identifiable reasons: a description written so the model cannot tell when the Skill applies, instructions that assume context the session will not have, or a scope wide enough that the Skill competes with the model’s own defaults. The session covers designing one Skill around a real use case, then building, testing, and refining it until it behaves consistently.
Included: the two-hour session, a step-by-step build, working test examples, slides, a recording, and a reusable design framework.
Early bird discount: Use code SKILLS19 to get it for 19$ until Tuesday 11 August 23:59 EEST, then $25, then $30.
Agent Skills for Claude Code Workshop
A large share of production LLM failures are not model failures. The model receives context that is incomplete, badly ordered, or padded with material competing with the part that mattered, and the output degrades in a way that looks like a capability problem.
Treating context as a system design problem makes selection, structure, and the assembly pipeline into components with their own failure modes — which can then be diagnosed rather than prompted around. The workshop covers those components, how context shapes output quality, how to diagnose weak or inconsistent results, current patterns, and how to design context pipelines for real agent workflows.
Included: the three-hour session, a pipeline design framework, worked examples, slides, recording, and take-home guidance.
Early bird discount: Use code CONTEXT35 to get it for $35 until Sunday 16 August 23:59 EEST, then $39, $45, $50.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.