diff --git a/mobile/src/components/pr-sidebar/markdown-blocks.ts b/mobile/src/components/pr-sidebar/markdown-blocks.ts
index d5866585b68..49d8e03e1f4 100644
--- a/mobile/src/components/pr-sidebar/markdown-blocks.ts
+++ b/mobile/src/components/pr-sidebar/markdown-blocks.ts
@@ -30,8 +30,6 @@ const HEADING = /^(#{1,6})\s+(.*)$/
 const FENCE = /^```/
 // Captures the fence info string (language) on the opening fence, e.g. ```mermaid.
 const FENCE_OPEN = /^```\s*([^\s`]*)/
-// A GFM table delimiter row: cells of dashes with optional leading/trailing colons.
-const TABLE_DELIM = /^\s*\|?\s*:?-{1,}:?\s*(\|\s*:?-{1,}:?\s*)*\|?\s*$/
 const QUOTE = /^>\s?(.*)$/
 const HR = /^(?:---+|\*\*\*+|___+)\s*$/
 const UNORDERED = /^\s*[-*+]\s+(.*)$/
@@ -114,7 +112,7 @@ function parseLines(content: string): MarkdownBlock[] {
 
     // GFM pipe table: a header row immediately followed by a delimiter row.
     // Requires the delimiter row so plain prose with a stray `|` isn't captured.
-    if (line.includes('|') && i + 1 < lines.length && TABLE_DELIM.test(lines[i + 1])) {
+    if (line.includes('|') && i + 1 < lines.length && isTableDelimiter(lines[i + 1])) {
       flushParagraph()
       const headers = splitTableRow(line)
       const align = parseAlignRow(lines[i + 1])
@@ -217,6 +215,10 @@ function splitTableRow(line: string): string[] {
   return cells
 }
 
+function isTableDelimiter(line: string): boolean {
+  return splitTableRow(line).every((cell) => /^:?-+:?$/.test(cell))
+}
+
 // Reads alignment from a delimiter row's colons: `:--` left, `:-:` center, `--:` right.
 function parseAlignRow(line: string): CellAlign[] {
   return splitTableRow(line).map((spec) => {
diff --git a/mobile/src/components/pr-sidebar/markdown-table-delimiter-performance.test.ts b/mobile/src/components/pr-sidebar/markdown-table-delimiter-performance.test.ts
new file mode 100644
index 00000000000..2f167feb747
--- /dev/null
+++ b/mobile/src/components/pr-sidebar/markdown-table-delimiter-performance.test.ts
@@ -0,0 +1,36 @@
+import { runInNewContext } from 'node:vm'
+import { describe, expect, it } from 'vitest'
+import { parseMarkdownBlocks } from './markdown-blocks'
+
+const ORIGINAL_DELIMITER = /^\s*\|?\s*:?-{1,}:?\s*(\|\s*:?-{1,}:?\s*)*\|?\s*$/
+
+function isParsedTable(delimiter: string): boolean {
+  return parseMarkdownBlocks(`a|b\n${delimiter}`)[0]?.kind === 'table'
+}
+
+describe('review Markdown table delimiter cost', () => {
+  it.each([
+    { name: 'leading whitespace', delimiter: ' '.repeat(60_000) + 'x' },
+    { name: 'trailing cell whitespace', delimiter: '-|-' + ' '.repeat(60_000) + 'x' }
+  ])('rejects $name without backtracking', ({ delimiter }) => {
+    // Interrupt synchronous regex regressions instead of hanging the test worker.
+    const blocks = runInNewContext(
+      'parse(input)',
+      { parse: parseMarkdownBlocks, input: `a|b\n${delimiter}` },
+      { timeout: 250 }
+    )
+    expect(blocks).toEqual([{ kind: 'paragraph', text: `a|b\n${delimiter}` }])
+  })
+
+  it('preserves the original delimiter grammar over generated rows', () => {
+    const parts = ['', '-', '--', ':', ':-:', ':--', '--:', '|', ' ', '\t', '\r', '\\|', 'x']
+    for (const left of parts) {
+      for (const middle of parts) {
+        for (const right of parts) {
+          const row = left + middle + right
+          expect(isParsedTable(row), JSON.stringify(row)).toBe(ORIGINAL_DELIMITER.test(row))
+        }
+      }
+    }
+  })
+})