KeKs0r

Codebase Research Benchmark: Standard vs Noodlbox

This benchmark compares two approaches to AI-powered codebase research:

• Standard (/cl:research_codebase) - Traditional file exploration using grep, glob, and file reads
• Noodlbox (/cl:research_codebase_noodl) - Knowledge graph-based exploration using Noodlbox

Both approaches were given identical questions about the Flick codebase (a Cloudflare-native error tracking system). The outputs were then evaluated by Claude on accuracy, completeness, actionability, and structure.

Results Summary

KeKs0r

import { subtle } from 'crypto'

import { Base64 } from 'js-base64'

type ServiceAccount = {
    private_key_id: string
    private_key: string
    client_email: string
}

KeKs0r

# Delete branches that were deleted on remote
alias git.clean_remote_deleted='git fetch --all -p; git branch -vv | grep ": gone]" | awk '"'"'{ print $1 }'"'"' | xargs -n 1 git branch -D'

# Delete branches that have been merged into main/dev/prod/master
alias git.clean_locally_merged='git branch --merged | egrep -v "(^\*|main|master|dev|prod)" | xargs git branch -d'

alias git.clean='git.clean_remote_deleted && git.clean_locally_merged'

KeKs0r

{
  "tailwindCSS.experimental.classRegex": [
    "tw:(.*)",
    "tw-start([\\w\\W]*?)tw-end",
    "[cC][lL][aA][sS][sS].+\"(.*)\"",
    "[cC][lL][aA][sS][sS].+'(.*)'",
    "clsx\\([\\w\\W]*?\\)",
    "twMerge\\([\\w\\W]*?\\)",
    "focus:(.*)",
    "hover:(.*)",

KeKs0r

How I currently write React Apps (2019 Hooks Edition)

Foreword

Most of my products are very MVP-ish, so I dont focus too much on super clean code and high quality, but there are a few patterns that I reuse quite a bit and find useful. But this is not an exhaustive list and the examples might not be as clean as one would think.

Technology Stack

Usually I am not using any sophisticated state Library anymore, the react hooks useState and useReducer in combination with context is absolutely sufficient, at least for UI-State. Everything else is in data persistence, which is in my case either Firebase or Graphql (= Apollo).

KeKs0r

// This actually shows the error in the zeit logs and results in this response
/*
An error occurred with this application.

NO_STATUS_CODE_FROM_LAMBDA
*/

async function start() {
  console.log("Before Server");
  const server = require("./server");

KeKs0r

import gql from "graphql-tag";
import { get } from "lodash";
import { useCallback } from "react";
import { useMutation } from "react-apollo-hooks";
import { GET_LEAD_HEADER } from "./getLeadHeader";

const CHANGE_STATUS = gql`
  mutation changeStatus($id: ID!, $status: String!) {
    changeStatus(id: $id, status: $status) {
      id

KeKs0r

const functions = require("firebase-functions");
const got = require("got");
const _ = require("lodash");

const notifyNewSignup = functions
  .runWith({ memory: "128MB" })
  .region("europe-west1")
  .auth.user()
  .onCreate(user => {
    const email = user.email;

KeKs0r

const Arena = require('bull-arena');
const _ = require('lodash')


function queueToConfig(queue){
    return {
        hostId: 'Main Host',
        name: queue.name,
        prefix: queue.keyPrefix,
        host: queue.eclient.options.host,

KeKs0r

function queryMiddleware() {
  return next => action => {
    if (action && action[SIMPLE] !== undefined && action[SIMPLE].query) {
      const request = action[SIMPLE];
      const query = qs.stringify(request.query, {
        arrayFormat: 'brackets',
        encode: false
      });
      request.endpoint = [request.endpoint.replace(/\?*/, ''), query].join('?');
      delete request.query;