diff --git a/src/scripts/analyse-nala.ts b/src/scripts/analyse-nala.ts
new file mode 100644
index 000000000..4bc1ae460
--- /dev/null
+++ b/src/scripts/analyse-nala.ts
@@ -0,0 +1,243 @@
+import { walk } from './common'
+import path from 'path'
+import fs from 'fs'
+import { execSync } from 'child_process'
+
+interface Options {
+  extractTimestamp: boolean
+}
+
+interface UsageMatch {
+  type: 'color' | 'spacing' | 'font'
+  raw: string
+  property: string
+  value: string
+  isNala: boolean
+  file: string
+  line: number
+  column: number
+  timestamp?: number
+}
+
+const allowedExtensions = [
+  '.html',
+  '.svelte',
+  '.css',
+  '.scss',
+  '.sass',
+  '.less',
+  '.lss',
+  '.js',
+  '.ts',
+  '.jsx',
+  '.tsx'
+]
+
+const spacingPatterns = [
+  /\spadding(-(top|right|bottom|left))?:/,
+  /\smargin(-(top|right|bottom|left))?:/,
+  /\s(gap|row-gap|column-gap):/,
+  /\s(top|right|bottom|left):/,
+  /\sinset(-(top|right|bottom|left))?:/
+]
+
+// There's a bunch of stuff we'll miss here (like setting just 'background')
+const colorPatterns = [
+  /\scolor:/,
+  /\sbackground-color:/,
+  /\sborder-color:/,
+  /\soutline-color:/,
+  /\sfill:/,
+  /\sstroke:/
+]
+
+const fontPatterns = [/\sfont((\w|-)+)?:/, /\sline((\w|-)+)?:/]
+
+const detectors = {
+  color: colorPatterns,
+  spacing: spacingPatterns,
+  font: fontPatterns
+}
+
+const isNalaValue = (value: string) => {
+  return (
+    value.includes('--leo-') ||
+    value.includes('color.') ||
+    value.includes('font.') ||
+    value.includes('spacing.')
+  )
+}
+
+const isFileTracked = (filepath: string) => {
+  // TODO: we shouldn't use try/catch here as it'll be slow. Instead we should
+  // check the exit code but I'm in a rush.
+  try {
+    execSync(`git ls-files --error-unmatch -- ${filepath}`, { stdio: 'ignore' })
+    return true
+  } catch (e) {
+    return false
+  }
+}
+
+const getLineDateLookup = (filepath: string): number[] | undefined => {
+  try {
+    const content = execSync(`git blame --line-porcelain -- ${filepath}`, {
+      encoding: 'utf8'
+    })
+
+    const lineDates: number[] = []
+
+    let lastTimeStamp = 0
+    for (const line of content.split('\n')) {
+      if (line.startsWith('author-time')) {
+        lastTimeStamp = parseInt(line.split(' ')[1])
+      } else if (line.startsWith('\t')) {
+        lineDates.push(lastTimeStamp)
+      }
+    }
+
+    return lineDates
+  } catch (e) {
+    return undefined
+  }
+}
+
+const walkAllowedFiles = () =>
+  walk(process.cwd(), (name, filepath, entry) => {
+    // Skip node_modules
+    if (name === 'node_modules') return true
+
+    // Skip type declarations
+    if (name.endsWith('.d.ts')) return true
+
+    // Skip SVGs
+    if (name.endsWith('.svg')) return true
+
+    // Skip test files
+    if (name.includes('.test.')) return true
+
+    // Skip vendor files
+    if (name.includes('vendor')) return true
+
+    // Skip third party files
+    if (name.includes('third_party')) return true
+
+    // If this isn't a directory or an allowed file extension, skip it
+    if (
+      !entry.isDirectory() &&
+      !allowedExtensions.includes(path.extname(filepath))
+    )
+      return true
+
+    // Skip untracked files
+    return !isFileTracked(filepath)
+  })
+
+function* getMatchesFromFile(
+  filepath: string,
+  options: Options
+): IterableIterator<UsageMatch> {
+  const content = fs.readFileSync(filepath, 'utf8')
+  const lineDates = options?.extractTimestamp
+    ? getLineDateLookup(filepath)
+    : undefined
+
+  let lineNumber = 0
+  const lines = content.split('\n')
+
+  for (const line of lines) {
+    lineNumber++
+    const getMatch = (type: 'color' | 'spacing', pattern: RegExpMatchArray) => {
+      const value = line
+        .substring(pattern.index! + pattern[0].length)
+        .trim()
+        .replace(';', '') // remove trailing semicolon, if any
+      return {
+        type: type,
+        raw: line.trim(),
+        property: pattern[0].trim().replace(':', ''), // remove trailing colon, if any
+        value,
+        isNala: isNalaValue(value),
+        file: filepath,
+        line: lineNumber,
+        column: pattern.index!,
+        timestamp: lineDates?.[lineNumber - 1]
+      }
+    }
+
+    let value: UsageMatch | null = null
+    for (const [type, patterns] of Object.entries(detectors)) {
+      const match = patterns
+        .map((pattern) => line.match(pattern))
+        .find((match) => match !== null)
+      if (match) {
+        value = getMatch(type as any, match)
+        break
+      }
+    }
+
+    // Check its not a JS thingy
+    // Note: These are heuristics and definitely will have false positives
+    // and false negatives.
+    const isJSThingy =
+      value?.value.startsWith('{') ||
+      value?.value.startsWith('"') ||
+      value?.value.startsWith("'") ||
+      value?.value.startsWith('`') ||
+      value?.value.startsWith('/') ||
+      line.includes(`let ${value?.value}`) ||
+      line.includes(`const ${value?.value}`) ||
+      line.includes(`var ${value?.value}`)
+
+    if (value && !isJSThingy) {
+      yield value
+    }
+  }
+}
+
+async function json() {
+  console.log('[')
+  for await (const filepath of await walkAllowedFiles()) {
+    for (const match of getMatchesFromFile(filepath, {
+      extractTimestamp: true
+    })) {
+      console.log(JSON.stringify(match, null, 4) + ',')
+    }
+  }
+  console.log(']')
+}
+
+async function csv() {
+  const maybeQuote = (v: any) => {
+    v = v.toString()
+    return v.includes(',')
+      ? `"${v.toString().replaceAll('"', "'")}"`
+      : v.replaceAll('"', "'")
+  }
+  console.log('type,property,value,isNala,file,line,column,timestamp,raw')
+  for await (const filepath of await walkAllowedFiles()) {
+    const relPath = path.relative(process.cwd(), filepath)
+    for (const match of getMatchesFromFile(filepath, {
+      extractTimestamp: true
+    })) {
+      const date = new Date(match.timestamp! * 1000)
+      console.log(
+        [
+          match.type,
+          match.property,
+          match.value,
+          match.isNala,
+          relPath,
+          match.line,
+          match.column,
+          `${date.getFullYear()}-${(date.getMonth() + 1).toString().padStart(2, '0')}-${date.getDay().toString().padStart(2, '0')}`,
+          match.raw
+        ]
+          .map(maybeQuote)
+          .join(',')
+      )
+    }
+  }
+}
+
+csv()