Cover Image for Claude Code Clinic
Cover Image for Claude Code Clinic
Avatar for Olinave
Presented by
Olinave
13 Went

Claude Code Clinic

Zoom
Registration
Past Event
Welcome! To join the event, please register below.
About Event

A **free, live online hour**: a focused, demo-driven teaching block followed by **open Q&A**.

An introduction to the Claude Code Control Surfaces framework for engineers, developers, architects or anyone using Claude Code regularly.

**The idea** — think of Claude Code like a car's dashboard:
**There are the controls** (your Control Surfaces — CLAUDE.md, skills.md, hooks, settings.json), **what they do** (how you actually steer a coding agent), and **the warning lights** (what to watch for if things are going wrong).

In this hour we **show you the controls and how to use them.** some prior knowledge of using Claude Code will be useful.

We anchor on the surface everyone knows - CLAUDE.md - place it in the context of our Control Surfaces map, then go deep on **three of the Olinave taxonomy's twelve failure modes**, focusing on the three most visible from CLAUDE.md:

**Compliance Illusion** — a plausible output or passing test doesn't prove your rules or instructions were followed.

- **Critical Rules in Prose** — a zero-tolerance "MUST" / "NEVER" written in prose reads as binding but only suggests what the agent should do.

- **Context / Attention Dilution** - an instruction fails not because it's absent, but because it's competing for attention with everything else in the agent's context.

**Who it's for** — anyone using Claude Code who's watched it ignore an instruction they cared about. No prep, no install needed: the live demo runs on a published repo, not yours.

**What you leave with** — one complete, proven "aha": why the agent ignores your CLAUDE.md, what it looks like when it happens, and the single move that fixes the worst cases.

**Credible** — a named taxonomy of the recurring failure modes, grounded in Olinave's controlled experiments and Anthropic's own published guidance, measured across **359** analysed Claude Code setups. Measurement method **patent-pending**. Taught by Philip Forshaw (Olinave; ex-Apple AI Strategy & Operations).

**It's free.** Want the full framework end to end? The Clinic ladders into the **Advanced Claude Code Masterclass** - same method, across all the surfaces, taught and demonstrated in a paid half-day session on July 23.

Avatar for Olinave
Presented by
Olinave
13 Went