by Anthropic
Built it with Claude Code?
Anthropic's agentic coding tool, run from the terminal or an editor. It works directly in your repository, reads files, runs commands, and makes multi-step changes.
What it generates
- Your existing stack, whatever it is
- Your repository and your infrastructure
- Project instruction files that carry conventions between sessions
- Can run tests and builds, and act on the results
What it’s good at
- Output tends to be well-structured and idiomatic, and it is comparatively good at following existing conventions in a codebase.
- Being able to run tests and read failures closes the loop on a lot of work without a person in the middle.
- Nothing is hosted or locked. It is your repository throughout.
- Genuinely capable of large, coherent changes rather than isolated snippets.
What Claude Code leaves you to build
Mostly scope rather than shortcoming — these are the boundaries of what the product sets out to do.
The architectural decisions
It will build well within a structure. Deciding what that structure should be depends on where the product is going, which is information that exists in your head and not in the repository.
Continuity across sessions
Each session starts without memory of the last. Instruction files carry conventions forward, but the reasoning behind past decisions is lost unless someone writes it down, and undocumented decisions get quietly reversed later.
Review that keeps pace with output
The constraint has moved. When a tool can produce a week of changes in an afternoon, review capacity becomes the bottleneck, and the temptation is to approve rather than to read.
Knowing when to stop
It is very good at doing what is asked. Whether the thing being asked for is worth building is a product judgement, and enthusiastic scope expansion is a common way for these projects to become unwieldy.
Signs you’ve outgrown it
- The project has grown past the point where you can hold its shape in your head.
- You are approving changes you have not fully read because there are too many of them.
- Sessions are re-litigating decisions that were settled weeks ago.
- Real users depend on it and nobody has ever reviewed the data-access layer specifically.
What a handoff involves
Claude Code — questions
My Claude Code project seems fine. Do I need a review?
It may well be fine, and we will say so if it is. The reason to check is that the categories of problem that persist are the ones that do not show up while using the app — access rules, authorization, and payment integrity all behave perfectly right up until someone deliberately probes them.
Do your engineers use Claude Code?
Yes, along with other tools. Declining to use them would make us slower without making the output better. What matters is that a person reviews, tests, and can explain everything that ships — and you should ask us to demonstrate that on any commit.
Last reviewed 2026-07-20. Claude Code ships changes frequently. If something here is out of date, tell us and we will correct it — we would rather be accurate than flattering.
Not sure what your Claude Code app needs?
A fixed-price review tells you what's actually wrong, what it would cost to fix, and whether you need us at all.