npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

@react-state-map/core

v0.3.2

Published

Static analysis engine for React: knows where every prop comes from. Import-aware component graph, prop tracing, impact analysis, prop-drilling and Next.js Server Component diagnostics, and safe lift-to-context refactors.

Readme

@react-state-map/core

The engine behind React State Map: a static analyzer that knows where every prop in a React app comes from.

Point it at a project and get a queryable graph of components, render edges, state, prop flows and contexts — resolved through imports exactly the way TypeScript does (path aliases, barrels, project references) — plus diagnostics, impact analysis and safe automatic refactors. It powers the VS Code extension and the CLI / MCP server; use it to build your own lint rules, codemods, dashboards or agent tools.

npm install @react-state-map/core

Node.js 18+. No runtime instrumentation and no build step: it reads your source files.

Quick start

import { ReactParser, GraphQuery } from '@react-state-map/core';

const parser = new ReactParser({ rootDir: './my-app' });
const { graph } = parser.parse();
const q = new GraphQuery(graph);

// Where does UserMenu's `onSelect` prop come from?
const menu = q.resolveComponent('UserMenu')!;
const trace = q.traceProp(menu.id, 'onSelect')!;
for (const origin of trace.origins) {
  console.log(origin.owner.name, origin.state.name, origin.chain.map(hop => hop.component.name));
  // → Shell selectedId [ 'Layout', 'Sidebar', 'UserMenu' ]
}

// What's affected if this state changes?
const state = q.findState('selectedId')[0]!;
console.log(q.impactOfState(state.id)!.summary);
// → 4 components affected (3 direct, 1 below)

// Everything worth fixing, with stable ids
for (const issue of graph.insights) {
  console.log(issue.severity, issue.code, `${issue.filePath}:${issue.line}`, issue.message);
}

What you get

ReactParser turns a project into a StateFlowGraph:

| Field | Contents | |---|---| | components | Every component (function, arrow, memo, forwardRef, class, anonymous default export) with props (types, usage), state it declares, contexts it provides/consumes, Server/Client environment | | renders | Parent → child render edges with the JSX call site and the props passed there (including Next.js layout → page) | | stateNodes | useState, useReducer, context, Redux, Zustand, Jotai, Recoil, Valtio, XState, TanStack Query, SWR, Apollo, React Hook Form, router hooks, custom hooks — with library and store/atom/query key | | edges | How each state value travels through props (with renames, spreads and setter tracking) and through context | | propDrillingPaths | Chains where a value passes through components that only forward it | | contextBoundaries, contextLeaks, bundles, componentMetrics, propChains | Context providers & consumers, context values re-passed as props, oversized object props, per-component prop usage roles, rename chains | | insights | Diagnostics with severity, location, related locations and a stable fingerprint id — see below | | meta | Files analyzed, duration, detected frameworks (next-app-router, react-19, …) |

Ids are stable and readable — c:src/components/Shell.tsx#Shell, s:src/components/Shell.tsx#Shell.selectedId, ctx:src/context/auth.tsx#AuthContext — so they survive edits and work as fingerprints in CI.

Insights

| Code | Severity | |---|---| | SERVER_COMPONENT_HOOK — client-only hook in a component rendered as a Server Component | error | | SERVER_TO_CLIENT_FUNCTION_PROP — function passed from a Server to a Client Component | error | | PROP_DRILLING — value passed through components that only forward it | warning | | CONTEXT_LEAK — context read and re-passed as props | info | | PROP_BUNDLE — large object passed as a single prop | info | | UNUSED_PROP — prop destructured but never used | hint | | PROP_PASSTHROUGH — one forwarding step of a drilling chain | hint |

Incremental analysis

The parser caches per-file facts. After the first parse(), feed it changes and parse again — only changed files are re-read, and the graph is re-linked in milliseconds. This is what keeps the editor integration and the MCP server live.

parser.updateFiles([
  { filePath: '/abs/src/Header.tsx', content: editorBuffer }, // unsaved text
  { filePath: '/abs/src/NewThing.tsx' },                       // created/changed on disk
  { filePath: '/abs/src/Old.tsx', content: null },             // deleted
]);
const next = parser.parse().graph;

Query API

new GraphQuery(graph) works on both the in-memory graph and its JSON form (serializeGraph / deserializeGraph). It only imports types, so you can load it on its own: import { GraphQuery } from '@react-state-map/core/query'.

| Method | | |---|---| | resolveComponent(nameOrId) / findComponents(query) / componentAt(file, line) | Lookup | | parents(id) / children(id) / descendants(id) / findRenderPath(from, to) | Render tree | | traceProp(componentId, prop) | Origin chains back to the owning state + every call site and the value passed there | | impactOfState(id) / impactOfComponent(id) / impactOfContext(id) | Affected components, grouped by reason | | contextConsumers(id) / contextProviders(id) / sharedSubscribers(state) | Who reads the same context / store / atom / query | | summarizeComponent(id) / drillingFor(id) / insightsInFile(file) | Per-component and per-file views |

Automatic fixes

Fix planners compute the edit in memory and return full before/after text per file — nothing is written, so you decide how to apply it.

import { planLiftToContext, planRemoveUnusedProp } from '@react-state-map/core';

const issue = graph.insights.find(i => i.code === 'PROP_DRILLING')!;
const plan = planLiftToContext(parser.getProject(), graph, issue.drillingPathId!, { react19: true });

if (plan.applicable) {
  for (const edit of plan.edits) {
    // edit.filePath, edit.oldText, edit.newText, edit.isNew
  }
} else {
  console.log(plan.reasons); // e.g. "Avatar is also rendered by Other; …"
}

planLiftToContext creates a typed context + hook next to the owner, provides it (memoized, <Ctx value> on React 19), removes the forwarded prop from every intermediate component's signature, types and JSX, and reads it with the hook in the consumer. It refuses — with reasons — whenever the change wouldn't be safe.

Text for humans and LLMs

@react-state-map/core/format holds the compact, path:line-annotated text formatters used by the MCP server and the Copilot tools (formatOverview, formatComponent, formatTraceProp, formatImpact, formatIssues, formatFixPlan, …), plus createUnifiedDiff and diffLineEdits for turning fix plans into patches or editor edits.

Options

new ReactParser({
  rootDir: './my-app',
  include: ['src/**/*.{ts,tsx}'],   // default: all .ts/.tsx/.js/.jsx/.mts/.mjs
  exclude: ['**/legacy/**'],         // default: node_modules, build output, tests, stories, .d.ts
  drillingThreshold: 3,              // hops before passing a value down counts as drilling
});

Module resolution follows the nearest tsconfig.json / jsconfig.json of each file (paths, baseUrl, extends, solution-style references), so monorepos with per-app configs just work.

License

MIT · GitHub · Issues