What a two-week Claude Expert Audit actually inspects

A Claude Expert Audit is two weeks inside your codebase and traces. The job is to find where quality, safety or money leaks, then hand back a ranked plan with baselines. It exists because most stalled pilots do not need a rewrite. They need a diagnosis. Background: why pilots never reach production.

The inspection list

1. Prompt and context. What is cached, what is retrieved, what is duplicated.
2. Tools. Schemas, permissions, error handling. See tool schema design.
3. Evals. Is there a golden set in CI, or a notebook? See eval harnesses.
4. Cost and latency. Spend per successful task, not per curiosity.
5. Safety. Injection, PII, review queues.
6. Operability. Tracing, owners, runbooks.

What you leave with

• A written baseline: pass rate, cost, latency on a held-out set
• A ranked remediation list
• A live readout with the people who own the system

You do not leave with a new platform or a hostage SaaS. The code stays in your repository.

When to audit versus rebuild

Audit when something already runs and you cannot say whether last week's change made it worse. Rebuild when there is no loop, no tools worth keeping, and no traces. We will tell you which one you are in. That honesty is the product.

Pricing starts at $9k for two weeks. Details live on engagements. If you want the list applied to your agent, book the fifteen-minute discussion.