diagnosing-react-code

Original：🇺🇸 English

Translated

Runs the react-doctor scanner on a React package, fixes its error diagnostics, and re-runs until none remain. Use when the user says "health-check the React code", or when React components or hooks changed, when finishing a React feature or bug fix, or when asked to health-check React code.

8installs

Sourcebostonaholic/skills

Added on2026-11-01

NPX Install

npx skill4agent add bostonaholic/skills diagnosing-react-code

SKILL.md Content

View Translation Comparison →

Diagnosing React code

react-doctor scans a React package and prints a 0-100 score plus diagnostics for security, performance, correctness, and architecture. It needs Node

^20.19.0 || >=22.13.0

.

Before the first run

The first run in a session downloads and executes a third-party package from npm. Ask the user before it, naming the pinned version.
By default the score comes from the react-doctor score API and the supply-chain scan queries Socket.dev; when the user wants nothing sent off the machine, add
```
--no-score --no-supply-chain
```
(the run then reports no score).

Run

Run inside the package whose

package.json

lists

react

in

dependencies

or

peerDependencies

; at a monorepo root the tool prompts for a project.

bash

npx -y react-doctor@0.9.17 --verbose --scope changed

The pin keeps scores comparable across runs. Never fall back to
```
@latest
```
; upgrade only by changing the pin deliberately. If a flag is rejected, check
```
npx -y react-doctor@0.9.17 --help
```
and adjust.
```
--verbose
```
lists every rule and file; the default shows only the top 3.
```
--scope changed
```
reports only new findings in files changed against the auto-detected base ref (
```
--base <ref>
```
to set it). Use
```
--scope full
```
for a whole-package health check. It replaces the deprecated
```
--diff
```
.

Fix loop

Record the baseline score and error and warning counts.
Fix error diagnostics, security and correctness first. Fix warnings only in code the current task changed.
Re-run the same command until it reports no errors and the score is at least the baseline. Stop after three fix rounds that leave errors, or when a fix would change behavior outside the task, and report what remains.

Report the version and scope, the score from baseline to final (or "not computed: --no-score"), each fixed diagnostic, and each remaining one with why it was not fixed, as

<severity> <rule> <file>:<line>

.