diff --git a/tools/cross/generate-compat-flag-docs.py b/tools/cross/generate-compat-flag-docs.py
new file mode 100755
index 00000000000..3d956e791a6
--- /dev/null
+++ b/tools/cross/generate-compat-flag-docs.py
@@ -0,0 +1,362 @@
+#!/usr/bin/env python3
+"""
+Tool to automatically generate documentation for compatibility flags.
+
+Reads compatibility flags from src/workerd/io/compatibility-date.capnp
+and generates markdown documentation files in the cloudflare-docs format.
+
+Skips:
+- Experimental flags (marked with $experimental)
+- Obsolete flags (field name starts with "obsolete")
+- Flags that already have documentation
+"""
+
+import re
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Optional
+
+import yaml
+
+
+@dataclass
+class CompatFlag:
+    """Represents a compatibility flag."""
+
+    field_name: str
+    enable_flag: Optional[str]
+    disable_flag: Optional[str]
+    enable_date: Optional[str]
+    implied_by_after_date: Optional[str]
+    description: str
+    is_experimental: bool
+    is_obsolete: bool
+
+    def get_doc_filename(self) -> Optional[str]:
+        """Generate the documentation filename from the enable flag name."""
+        if not self.enable_flag:
+            return None
+        # Convert flag name to kebab-case filename
+        return f"{self.enable_flag.replace('_', '-')}.md"
+
+    def get_human_readable_name(self) -> str:
+        """Generate a human-readable name from the enable flag."""
+        if not self.enable_flag:
+            return self.field_name
+        # Convert snake_case to Title Case
+        words = self.enable_flag.replace("_", " ").split()
+        return " ".join(word.capitalize() for word in words)
+
+    def should_generate_docs(self, documented_flags: set[str]) -> bool:
+        """Check if documentation should be generated for this flag."""
+        if self.is_experimental:
+            return False
+        if self.is_obsolete:
+            return False
+        if not self.enable_flag:
+            return False
+
+        # Skip if flag has neither an enable date nor an implied by after date
+        if not self.enable_date and not self.implied_by_after_date:
+            return False
+
+        # Check if this flag is already documented
+        if self.enable_flag in documented_flags:
+            return False
+
+        return True
+
+
+def parse_capnp_file(filepath: Path) -> list[CompatFlag]:
+    """Parse the compatibility-date.capnp file and extract flags."""
+    content = filepath.read_text()
+
+    flags = []
+
+    # Find all field definitions
+    # Pattern: field_name @number :Bool followed by annotations and comments
+    pattern = r"(\w+)\s+@\d+\s*:Bool\s*([^;]*);"
+
+    matches = re.finditer(pattern, content, re.MULTILINE | re.DOTALL)
+
+    for match in matches:
+        field_name = match.group(1)
+        annotations_block = match.group(2)
+
+        # Check if obsolete
+        is_obsolete = field_name.startswith("obsolete")
+
+        # Extract annotations
+        enable_flag = None
+        disable_flag = None
+        enable_date = None
+        implied_by_after_date = None
+        is_experimental = False
+
+        # Extract enable flag
+        enable_match = re.search(r'\$compatEnableFlag\("([^"]+)"\)', annotations_block)
+        if enable_match:
+            enable_flag = enable_match.group(1)
+
+        # Extract disable flag
+        disable_match = re.search(
+            r'\$compatDisableFlag\("([^"]+)"\)', annotations_block
+        )
+        if disable_match:
+            disable_flag = disable_match.group(1)
+
+        # Extract enable date
+        date_match = re.search(r'\$compatEnableDate\("([^"]+)"\)', annotations_block)
+        if date_match:
+            enable_date = date_match.group(1)
+
+        # Extract implied by after date
+        implied_match = re.search(
+            r'\$impliedByAfterDate\([^)]*date\s*=\s*"([^"]+)"', annotations_block
+        )
+        if implied_match:
+            implied_by_after_date = implied_match.group(1)
+
+        # Check if experimental
+        if "$experimental" in annotations_block:
+            is_experimental = True
+
+        # Extract description from comments after the field
+        # Find the position after this field's semicolon
+        field_end = match.end()
+
+        # Look forward for comments
+        after_field = content[field_end:]
+        lines_after = after_field.split("\n")
+
+        # Collect comment lines
+        description_lines = []
+        for line in lines_after:
+            stripped = line.strip()
+            if stripped.startswith("#"):
+                # Remove the leading # and whitespace
+                comment_text = stripped[1:].strip()
+                description_lines.append(comment_text)
+            elif stripped == "":
+                # Empty line - stop if we've already collected comments
+                if description_lines:
+                    break
+                # Otherwise continue looking
+                continue
+            else:
+                # Non-comment, non-empty line - stop
+                break
+
+        description = " ".join(description_lines) if description_lines else ""
+
+        flag = CompatFlag(
+            field_name=field_name,
+            enable_flag=enable_flag,
+            disable_flag=disable_flag,
+            enable_date=enable_date,
+            implied_by_after_date=implied_by_after_date,
+            description=description,
+            is_experimental=is_experimental,
+            is_obsolete=is_obsolete,
+        )
+
+        flags.append(flag)
+
+    return flags
+
+
+def parse_frontmatter(filepath: Path) -> dict[str, str]:
+    """Parse the YAML frontmatter from a markdown file."""
+    try:
+        content = filepath.read_text()
+
+        # Extract frontmatter between --- markers
+        match = re.match(r"^---\s*\n(.*?)\n---\s*\n", content, re.DOTALL)
+        if not match:
+            return {}
+
+        frontmatter_text = match.group(1)
+
+        # Parse YAML
+        try:
+            frontmatter = yaml.safe_load(frontmatter_text)
+            return frontmatter if isinstance(frontmatter, dict) else {}
+        except yaml.YAMLError:
+            return {}
+    except Exception:
+        return {}
+
+
+def get_existing_docs(docs_dir: str) -> set[str]:
+    """
+    Get the set of compatibility flags that are already documented.
+
+    Returns a set of enable_flag and disable_flag values found in existing docs.
+    """
+    docs_path = Path(docs_dir)
+    if not docs_path.exists():
+        return set()
+
+    documented_flags = set()
+
+    for filepath in docs_path.iterdir():
+        if not filepath.is_file() or filepath.suffix != ".md":
+            continue
+
+        frontmatter = parse_frontmatter(filepath)
+
+        # Add both enable_flag and disable_flag if present
+        if "enable_flag" in frontmatter:
+            documented_flags.add(frontmatter["enable_flag"])
+        if "disable_flag" in frontmatter:
+            documented_flags.add(frontmatter["disable_flag"])
+
+    return documented_flags
+
+
+def generate_markdown(flag: CompatFlag) -> str:
+    """Generate markdown documentation for a flag."""
+
+    # Prepare the frontmatter
+    frontmatter = [
+        "---",
+        "_build:",
+        "  publishResources: false",
+        "  render: never",
+        "  list: never",
+        "",
+        f'name: "{flag.get_human_readable_name()}"',
+    ]
+
+    if flag.enable_date:
+        frontmatter.append(f'sort_date: "{flag.enable_date}"')
+
+    if flag.enable_flag:
+        frontmatter.append(f'enable_flag: "{flag.enable_flag}"')
+
+    if flag.disable_flag:
+        frontmatter.append(f'disable_flag: "{flag.disable_flag}"')
+
+    frontmatter.append("---")
+
+    # Prepare the body
+    body = []
+
+    if flag.description:
+        body.append(flag.description)
+    else:
+        body.append(f"Compatibility flag: `{flag.enable_flag}`")
+
+    # Combine everything
+    return "\n".join(frontmatter) + "\n\n" + "\n".join(body) + "\n"
+
+
+def main():
+    """Main entry point."""
+    import argparse
+
+    parser = argparse.ArgumentParser(
+        description="Generate compatibility flag documentation"
+    )
+    parser.add_argument(
+        "--capnp-file",
+        default="src/workerd/io/compatibility-date.capnp",
+        help="Path to compatibility-date.capnp file",
+        type=Path,
+    )
+    parser.add_argument(
+        "--docs-dir",
+        default="../cloudflare-docs/src/content/compatibility-flags",
+        help="Path to cloudflare-docs compatibility-flags directory",
+        type=Path,
+    )
+    parser.add_argument(
+        "--output-dir", help="Output directory (defaults to --docs-dir)", type=Path
+    )
+    parser.add_argument(
+        "--dry-run",
+        action="store_true",
+        help="Print what would be generated without writing files",
+    )
+    parser.add_argument(
+        "--force", action="store_true", help="Generate docs even if they already exist"
+    )
+    parser.add_argument(
+        "--verbose",
+        action="store_true",
+        help="Show detailed information about flag detection",
+    )
+
+    args = parser.parse_args()
+
+    output_dir = args.output_dir or args.docs_dir
+
+    # Parse the capnp file
+    print(f"Parsing {args.capnp_file}...")
+    flags = parse_capnp_file(args.capnp_file)
+    print(f"Found {len(flags)} total flags")
+
+    # Get existing docs
+    documented_flags = get_existing_docs(args.docs_dir) if not args.force else set()
+    print(f"Found {len(documented_flags)} already-documented compatibility flags")
+
+    if args.verbose and documented_flags:
+        print("\nAlready documented flags:")
+        for flag_name in sorted(documented_flags):
+            print(f"  - {flag_name}")
+
+    # Filter flags that need documentation
+    flags_to_document = [
+        f for f in flags if f.should_generate_docs(documented_flags) or args.force
+    ]
+
+    print(f"\nFlags to document: {len(flags_to_document)}")
+    print(f"Skipped experimental: {sum(1 for f in flags if f.is_experimental)}")
+    print(f"Skipped obsolete: {sum(1 for f in flags if f.is_obsolete)}")
+    print(
+        f"Skipped without dates: {sum(1 for f in flags if not f.is_experimental and not f.is_obsolete and f.enable_flag and not f.enable_date and not f.implied_by_after_date)}"
+    )
+    print(
+        f"Already documented: {sum(1 for f in flags if not f.is_experimental and not f.is_obsolete and f.enable_flag and f.enable_flag in documented_flags)}"
+    )
+    print(
+        f"Missing enable flag: {sum(1 for f in flags if not f.enable_flag and not f.is_obsolete)}"
+    )
+
+    if not flags_to_document:
+        print("\nNo flags to document!")
+        return
+
+    # Generate documentation
+    print(f"\n{'Would generate' if args.dry_run else 'Generating'} documentation for:")
+
+    for flag in flags_to_document:
+        filename = flag.get_doc_filename()
+        if not filename:
+            continue
+
+        filepath = output_dir / filename
+        markdown = generate_markdown(flag)
+
+        print(f"  - {filename}")
+
+        if args.dry_run:
+            print(f"\n--- {filename} ---")
+            print(markdown)
+            print("--- End ---\n")
+        else:
+            output_dir.mkdir(parents=True, exist_ok=True)
+            filepath.write_text(markdown)
+
+    if not args.dry_run:
+        print(
+            f"\nGenerated {len(flags_to_document)} documentation files in {output_dir}"
+        )
+    else:
+        print(
+            f"\nDry run complete. Would have generated {len(flags_to_document)} files."
+        )
+
+
+if __name__ == "__main__":
+    main()