anonymous's avatar

@anonymous's Claude Code Insights

2026-05-25 — 2026-10-07 | 494 messages | 98 sessions

494 Messages
+28K/-1K Lines Changed
333 Files
18.3 Msgs/Day

At a Glance

What's working: You have a consistent habit of asking for an impact analysis or plan before any code is touched, which paid off when the new ACU-Serve format landed as an isolated strategy without breaking Shopify or WordPress modes. You also took the dome-chatbox component from no tests to a layered, verified suite through a spec-driven flow, narrowing the scope first so it stayed manageable. On layout work, you give concrete visual feedback and close out with a regression check and a convention-following commit message.
What's hindering: On Claude's side, first implementations often passed type-check and lint but had runtime bugs you ended up finding yourself (a save that returned an error, a toggle that did nothing, excess empty space in a layout). Claude also misread the deliverable several times, explaining when you wanted an edit, following the spec flow when you wanted direct implementation, or writing a plan in English when you were writing in Spanish. On your side, many planning sessions cycled through several rejected plans, and some analyses were restarted from scratch in new sessions; scope, language and whether you want a plan at all were usually not stated up front, and expired tokens or missing local runtimes blocked a few sessions.

Big Wins

Impact analysis before implementation

You routinely ask for an impact analysis or plan before any code is touched, as with the game-level unification, the assets[] format and the SurveyMonkey script swap. That gives you code-referenced findings across frontend, backend and routing, and it paid off when the ACU-Serve format was added as an isolated strategy with Shopify and WordPress regression cases checked.

Spec-driven test suite building

You took the dome-chatbox component from zero tests to a layered suite through an approved spec, plan and tasks flow. The result was 97 unit/DOM tests plus 3 Playwright E2E tests, all verified, and you narrowed the spec scope to a single component first so the work stayed manageable.

Layout fixes with tight feedback

On the split-flow zigzag work you reported concrete visual problems, like the excess empty space and the missing alternation, and got targeted fixes in the renderer and CSS. You also close out changes consistently by asking whether Shopify mode is affected and by requesting a branch name and commit message that follow the repo convention.

Top Tools

Languages

Session Types