diff --git a/.github/workflows/accessibility-regression.yml b/.github/workflows/accessibility-regression.yml
index 003a483..6d2bd7d 100644
--- a/.github/workflows/accessibility-regression.yml
+++ b/.github/workflows/accessibility-regression.yml
@@ -56,6 +56,7 @@ jobs:
       - name: Install Python dependencies
         run: |
           python -m pip install --upgrade pip
+          python -m pip install --upgrade --force-reinstall "git+https://github.com/Community-Access/quill-glow-core.git@main"
           python -m pip install -e ./desktop
           python -m pip install -e ./web
 
diff --git a/.github/workflows/deploy.yml b/.github/workflows/deploy.yml
index e6d37a6..a51ec6c 100644
--- a/.github/workflows/deploy.yml
+++ b/.github/workflows/deploy.yml
@@ -60,6 +60,7 @@ jobs:
             done
           }
 
+          install_with_retry 4 python -m pip install --upgrade --force-reinstall "git+https://github.com/Community-Access/quill-glow-core.git@main"
           install_with_retry 4 python -m pip install -e ./desktop
           install_with_retry 4 python -m pip install -e ./web
           install_with_retry 4 python -m pip install -r ./mcp_server/requirements.txt
diff --git a/.github/workflows/feature-flags-ci.yml b/.github/workflows/feature-flags-ci.yml
index c0dd6bb..9979437 100644
--- a/.github/workflows/feature-flags-ci.yml
+++ b/.github/workflows/feature-flags-ci.yml
@@ -39,14 +39,17 @@ jobs:
       - name: Install dependencies (desktop and web dev)
         run: |
           python -m pip install --upgrade pip setuptools wheel
+          python -m pip install --upgrade --force-reinstall "git+https://github.com/Community-Access/quill-glow-core.git@main"
           python -m pip install -e desktop/
           python -m pip install -e web[dev]
 
       - name: Run feature flags migration
         run: python web/tools/run_migrate.py
 
-      - name: Run web tests
-        run: python -m pytest web/ -q
+      - name: Run focused feature-flag regression tests
+        run: |
+          python -m pytest web/tests/test_admin_flags.py -q
+          python -m pytest web/tests/test_ai_feature_gates.py -q
         env:
           # ensure sqlite backend for audit tests
           FEATURE_FLAGS_BACKEND: sqlite
diff --git a/README.md b/README.md
index e78640d..2e00555 100644
--- a/README.md
+++ b/README.md
@@ -124,6 +124,27 @@ Current rules are intentionally low-noise and focus on:
 - Produces PowerShell scripts for configuring Word document styles
 - Detects and uses external tools (markdownlint, Pandoc) when available
 
+## Shared core architecture (retrofit baseline)
+
+GLOW now includes a shared service package at:
+
+- `desktop/src/acb_large_print_core/`
+
+This package is the canonical dispatch layer for **audit**, **fix**, and **MarkItDown conversion** across CLI, desktop, and web surfaces.
+
+Current shared entry points:
+
+- `acb_large_print_core.services.audit_by_extension(...)`
+- `acb_large_print_core.services.fix_by_extension(...)`
+- `acb_large_print_core.services.convert_to_markdown(...)`
+- `acb_large_print_core.versions.get_component_versions()`
+
+Design intent:
+
+- Keep UX-specific behavior in each interface (web routes, GUI dialogs, CLI output formatting).
+- Keep business logic dispatch and component-version provenance in shared core APIs.
+- Enable future extraction to an external reusable package without changing app-layer behavior.
+
 ## Recent Fix Workflow Updates (April 2026)
 
 - Fix Results now suppresses `ACB-FAUX-HEADING` from post-fix scoring when heading detection is explicitly disabled, and shows a "Suppressed by your settings" note for transparency.
@@ -169,6 +190,7 @@ lp/
     announcement.md                      Press release / announcement
     prd.md                               Canonical web app product requirements document
     deployment.md                        Step-by-step server deployment guide
+    shared-core-retrofit.md              Shared service-core architecture and migration status
   samples/
     *.md                                 Example Markdown source files
     *.html                               Converted HTML output files
@@ -181,7 +203,8 @@ lp/
     Dockerfile                           Production container image
     docker-compose.yml                   Compose file for deployment
   desktop/                               Desktop CLI + GUI (Python)
-    src/acb_large_print/                 Core library (canonical source of truth)
+    src/acb_large_print_core/            Shared service core (canonical dispatch layer)
+    src/acb_large_print/                 App-layer CLI/GUI + format engines
   office-addin/                          Office.js Word Add-in (TypeScript)
     src/                                 TypeScript port of audit/fix/template
   vendor/                                Vendored third-party source
diff --git a/desktop/pyproject.toml b/desktop/pyproject.toml
index 43d1304..c600c66 100644
--- a/desktop/pyproject.toml
+++ b/desktop/pyproject.toml
@@ -28,6 +28,7 @@ classifiers = [
     "Topic :: Adaptive Technologies",
 ]
 dependencies = [
+    "quill-glow-core>=0.1.0",
     "python-docx>=1.2.0",
     "mammoth>=1.11.0,<1.12.0",
     "markitdown[all]>=0.1.5",
diff --git a/desktop/requirements.txt b/desktop/requirements.txt
index aa37bd0..2ec7294 100644
--- a/desktop/requirements.txt
+++ b/desktop/requirements.txt
@@ -1,4 +1,5 @@
 # Core dependencies
+quill-glow-core>=0.1.0
 python-docx>=1.1.0
 mammoth>=1.11.0,<1.12.0
 markitdown[all]>=0.1.5
diff --git a/desktop/src/acb_large_print/__init__.py b/desktop/src/acb_large_print/__init__.py
index fa34ddc..27ccdbd 100644
--- a/desktop/src/acb_large_print/__init__.py
+++ b/desktop/src/acb_large_print/__init__.py
@@ -1,5 +1,13 @@
-"""ACB Large Print Tool -- audit, fix, and template Word documents for ACB compliance."""
+"""ACB Large Print Tool -- audit, fix, and convert accessible documents."""
+
+from __future__ import annotations
 
-__version__ = "5.0.0"
 __app_name__ = "ACB Large Print Tool"
 __author__ = "BITS (Blind Information Technology Solutions)"
+
+try:
+    from .version import get_version as _get_version
+
+    __version__ = _get_version()
+except Exception:
+    __version__ = "unknown"
diff --git a/desktop/src/acb_large_print/cli.py b/desktop/src/acb_large_print/cli.py
index fc3464c..1e63c4a 100644
--- a/desktop/src/acb_large_print/cli.py
+++ b/desktop/src/acb_large_print/cli.py
@@ -713,28 +713,14 @@ def _build_parser() -> argparse.ArgumentParser:
 
 
 # ── Supported file extensions ─────────────────────────────────────────
-SUPPORTED_EXTENSIONS = {".docx", ".xlsx", ".pptx", ".epub"}
+SUPPORTED_EXTENSIONS = {".docx", ".xlsx", ".pptx", ".md", ".pdf", ".epub"}
 
 
 def _audit_by_extension(file_path: Path):
     """Dispatch to the correct auditor based on file extension."""
-    ext = file_path.suffix.lower()
-    if ext == ".xlsx":
-        from .xlsx_auditor import audit_workbook
+    from quill_glow_core import audit_by_extension
 
-        return audit_workbook(file_path)
-    elif ext == ".pptx":
-        from .pptx_auditor import audit_presentation
-
-        return audit_presentation(file_path)
-    elif ext == ".epub":
-        from .epub_auditor import audit_epub
-
-        return audit_epub(file_path)
-    else:
-        from .auditor import audit_document
-
-        return audit_document(file_path)
+    return audit_by_extension(file_path)
 
 
 def _resolve_list_indent(args: argparse.Namespace) -> tuple[float, float]:
@@ -819,61 +805,17 @@ def _fix_by_extension(
 
     Returns (output_path, total_fixes, fix_records, post_audit, warnings).
     """
-    ext = file_path.suffix.lower()
-    if ext == ".xlsx":
-        from .xlsx_auditor import audit_workbook
-
-        post_audit = audit_workbook(file_path)
-        return (
-            file_path,
-            0,
-            [],
-            post_audit,
-            [
-                "Excel workbooks cannot be auto-fixed yet. "
-                "Review the audit findings and fix them manually in Excel."
-            ],
-        )
-    elif ext == ".pptx":
-        from .pptx_auditor import audit_presentation
-
-        post_audit = audit_presentation(file_path)
-        return (
-            file_path,
-            0,
-            [],
-            post_audit,
-            [
-                "PowerPoint presentations cannot be auto-fixed yet. "
-                "Review the audit findings and fix them manually in PowerPoint."
-            ],
-        )
-    elif ext == ".epub":
-        from .epub_auditor import audit_epub
+    from quill_glow_core import fix_by_extension
 
-        post_audit = audit_epub(file_path)
-        return (
-            file_path,
-            0,
-            [],
-            post_audit,
-            [
-                "ePub files cannot be auto-fixed yet. "
-                "Review the audit findings and fix them in your ePub editor."
-            ],
-        )
-    else:
-        from .fixer import fix_document
-
-        return fix_document(
-            file_path,
-            output_path=output_path,
-            bound=bound,
-            list_indent_in=list_indent_in,
-            list_hanging_in=list_hanging_in,
-            para_indent_in=para_indent_in,
-            first_line_indent_in=first_line_indent_in,
-        )
+    return fix_by_extension(
+        file_path,
+        output_path=output_path,
+        bound=bound,
+        list_indent_in=list_indent_in,
+        list_hanging_in=list_hanging_in,
+        para_indent_in=para_indent_in,
+        first_line_indent_in=first_line_indent_in,
+    )
 
 
 def _cmd_audit(args: argparse.Namespace) -> int:
@@ -885,9 +827,10 @@ def _cmd_audit(args: argparse.Namespace) -> int:
         return 1
 
     ext = args.file.suffix.lower()
-    if ext not in (".docx", ".xlsx", ".pptx", ".epub"):
+    if ext not in SUPPORTED_EXTENSIONS:
         print(
-            f"Error: Unsupported file type '{ext}'. Use .docx, .xlsx, .pptx, or .epub.",
+            "Error: Unsupported file type "
+            f"'{ext}'. Use .docx, .xlsx, .pptx, .md, .pdf, or .epub.",
             file=sys.stderr,
         )
         return 1
@@ -927,7 +870,8 @@ def _cmd_fix(args: argparse.Namespace) -> int:
     ext = args.file.suffix.lower()
     if ext not in SUPPORTED_EXTENSIONS:
         print(
-            f"Error: Unsupported file type '{ext}'. Use .docx, .xlsx, .pptx, or .epub.",
+            "Error: Unsupported file type "
+            f"'{ext}'. Use .docx, .xlsx, .pptx, .md, .pdf, or .epub.",
             file=sys.stderr,
         )
         return 1
@@ -1291,7 +1235,8 @@ def _print_wcag_language_report(report) -> None:
 
 def _cmd_convert(args: argparse.Namespace) -> int:
     """Execute the convert command."""
-    from .converter import CONVERTIBLE_EXTENSIONS, convert_to_markdown
+    from .converter import CONVERTIBLE_EXTENSIONS
+    from quill_glow_core import convert_to_markdown
     from .wcag_language import analyze_text_for_wcag_language
 
     if not args.file.exists():
diff --git a/desktop/src/acb_large_print/cli_main.py b/desktop/src/acb_large_print/cli_main.py
index 2bc5ff3..a0dacad 100644
--- a/desktop/src/acb_large_print/cli_main.py
+++ b/desktop/src/acb_large_print/cli_main.py
@@ -6,6 +6,13 @@
 
 
 def main() -> None:
+    try:
+        from quill_glow_core import configure_default_services as _configure_shared_core_default
+
+        _configure_shared_core_default()
+    except Exception:
+        pass
+
     from acb_large_print.cli import main as cli_main
 
     sys.exit(cli_main(force_cli=True))
diff --git a/desktop/src/acb_large_print_core/__init__.py b/desktop/src/acb_large_print_core/__init__.py
new file mode 100644
index 0000000..c99c73a
--- /dev/null
+++ b/desktop/src/acb_large_print_core/__init__.py
@@ -0,0 +1,23 @@
+"""Shared core services for GLOW audit/fix/convert workflows."""
+
+from .services import (
+    CONVERTIBLE_EXTENSIONS,
+    MARKITDOWN_AUDIO_EXTENSIONS,
+    SUPPORTED_AUDIT_EXTENSIONS,
+    SUPPORTED_FIX_EXTENSIONS,
+    audit_by_extension,
+    convert_to_markdown,
+    fix_by_extension,
+)
+from .versions import get_component_versions
+
+__all__ = [
+    "CONVERTIBLE_EXTENSIONS",
+    "MARKITDOWN_AUDIO_EXTENSIONS",
+    "SUPPORTED_AUDIT_EXTENSIONS",
+    "SUPPORTED_FIX_EXTENSIONS",
+    "audit_by_extension",
+    "convert_to_markdown",
+    "fix_by_extension",
+    "get_component_versions",
+]
diff --git a/desktop/src/acb_large_print_core/services.py b/desktop/src/acb_large_print_core/services.py
new file mode 100644
index 0000000..abcd96d
--- /dev/null
+++ b/desktop/src/acb_large_print_core/services.py
@@ -0,0 +1,173 @@
+"""Canonical shared dispatch services for GLOW desktop/web/CLI."""
+
+from __future__ import annotations
+
+from pathlib import Path
+
+SUPPORTED_AUDIT_EXTENSIONS = {".docx", ".xlsx", ".pptx", ".md", ".pdf", ".epub"}
+SUPPORTED_FIX_EXTENSIONS = set(SUPPORTED_AUDIT_EXTENSIONS)
+
+
+def audit_by_extension(
+    file_path: str | Path,
+    *,
+    list_indent_in: float | None = None,
+    list_level_indents: dict[int, float] | None = None,
+    para_indent_in: float | None = None,
+    first_line_indent_in: float | None = None,
+    style_size_overrides: dict[str, float] | None = None,
+):
+    """Run the appropriate auditor for the file extension."""
+    path = Path(file_path)
+    ext = path.suffix.lower()
+
+    if ext == ".xlsx":
+        from acb_large_print.xlsx_auditor import audit_workbook
+
+        return audit_workbook(path)
+    if ext == ".pptx":
+        from acb_large_print.pptx_auditor import audit_presentation
+
+        return audit_presentation(path)
+    if ext == ".md":
+        from acb_large_print.md_auditor import audit_markdown
+
+        return audit_markdown(path)
+    if ext == ".pdf":
+        from acb_large_print.pdf_auditor import audit_pdf
+
+        return audit_pdf(path)
+    if ext == ".epub":
+        from acb_large_print.epub_auditor import audit_epub
+
+        return audit_epub(path)
+
+    from acb_large_print.auditor import audit_document
+
+    return audit_document(
+        path,
+        list_indent_in=list_indent_in,
+        list_level_indents=list_level_indents,
+        para_indent_in=para_indent_in,
+        first_line_indent_in=first_line_indent_in,
+        style_size_overrides=style_size_overrides,
+    )
+
+
+def fix_by_extension(
+    file_path: str | Path,
+    output_path: str | Path | None = None,
+    *,
+    bound: bool = False,
+    list_indent_in: float = 0.0,
+    list_hanging_in: float = 0.0,
+    list_level_indents: dict[int, float] | None = None,
+    para_indent_in: float = 0.0,
+    first_line_indent_in: float = 0.0,
+    preserve_heading_alignment: bool = False,
+    detect_headings: bool = False,
+    ai_provider: object | None = None,
+    heading_threshold: int | None = None,
+    confirmed_headings: list | None = None,
+    heading_accuracy_level: str = "balanced",
+    style_size_overrides: dict[str, float] | None = None,
+):
+    """Run fixer workflow for the extension.
+
+    Returns (output_path, total_fixes, fix_records, post_audit, warnings).
+    """
+    path = Path(file_path)
+    out = Path(output_path) if output_path is not None else None
+    ext = path.suffix.lower()
+
+    if ext == ".xlsx":
+        post_audit = audit_by_extension(path)
+        return (
+            path,
+            0,
+            [],
+            post_audit,
+            [
+                "Excel workbooks cannot be auto-fixed yet. "
+                "Review the audit findings and fix them manually in Excel."
+            ],
+        )
+    if ext == ".pptx":
+        post_audit = audit_by_extension(path)
+        return (
+            path,
+            0,
+            [],
+            post_audit,
+            [
+                "PowerPoint presentations cannot be auto-fixed yet. "
+                "Review the audit findings and fix them manually in PowerPoint."
+            ],
+        )
+    if ext == ".md":
+        post_audit = audit_by_extension(path)
+        return (
+            path,
+            0,
+            [],
+            post_audit,
+            [
+                "Markdown auto-fix is coming soon. "
+                "Review the audit findings and fix them in your text editor."
+            ],
+        )
+    if ext == ".pdf":
+        post_audit = audit_by_extension(path)
+        return (
+            path,
+            0,
+            [],
+            post_audit,
+            [
+                "PDF files cannot be auto-fixed. "
+                "Use Adobe Acrobat Pro or re-export from the source application."
+            ],
+        )
+    if ext == ".epub":
+        post_audit = audit_by_extension(path)
+        return (
+            path,
+            0,
+            [],
+            post_audit,
+            [
+                "ePub files cannot be auto-fixed yet. "
+                "Review the audit findings and fix them in your ePub editor."
+            ],
+        )
+
+    from acb_large_print.fixer import fix_document
+
+    return fix_document(
+        path,
+        output_path=out,
+        bound=bound,
+        list_indent_in=list_indent_in,
+        list_hanging_in=list_hanging_in,
+        list_level_indents=list_level_indents,
+        para_indent_in=para_indent_in,
+        first_line_indent_in=first_line_indent_in,
+        preserve_heading_alignment=preserve_heading_alignment,
+        detect_headings=detect_headings,
+        ai_provider=ai_provider,
+        heading_threshold=heading_threshold,
+        confirmed_headings=confirmed_headings,
+        heading_accuracy_level=heading_accuracy_level,
+        style_size_overrides=style_size_overrides,
+    )
+
+
+from acb_large_print import converter as _converter  # noqa: E402
+
+CONVERTIBLE_EXTENSIONS = _converter.CONVERTIBLE_EXTENSIONS
+MARKITDOWN_AUDIO_EXTENSIONS = _converter.MARKITDOWN_AUDIO_EXTENSIONS
+
+
+def convert_to_markdown(src_path: str | Path, output_path: str | Path | None = None):
+    """Convert source document to Markdown via shared MarkItDown pipeline."""
+    return _converter.convert_to_markdown(src_path, output_path=output_path)
diff --git a/desktop/src/acb_large_print_core/versions.py b/desktop/src/acb_large_print_core/versions.py
new file mode 100644
index 0000000..460f8bf
--- /dev/null
+++ b/desktop/src/acb_large_print_core/versions.py
@@ -0,0 +1,42 @@
+"""Component version manifest for support and diagnostics."""
+
+from __future__ import annotations
+
+from importlib.metadata import PackageNotFoundError, version
+
+
+def _pkg_version(name: str) -> str:
+    try:
+        return version(name)
+    except PackageNotFoundError:
+        return "not-installed"
+
+
+def _module_version(module_name: str) -> str:
+    try:
+        module = __import__(module_name)
+    except Exception:
+        return "not-installed"
+    return str(getattr(module, "__version__", "unknown"))
+
+
+def get_component_versions() -> dict[str, str]:
+    """Return a normalized component version map used by desktop/web surfaces."""
+    try:
+        from acb_large_print.version import get_version as _get_release_version
+
+        release_version = _get_release_version()
+    except Exception:
+        release_version = _pkg_version("acb-large-print")
+
+    return {
+        "release_version": release_version,
+        "desktop_package": _pkg_version("acb-large-print"),
+        "web_package": _pkg_version("acb-large-print-web"),
+        "markitdown": _pkg_version("markitdown"),
+        "pymupdf": _module_version("fitz"),
+        "python_docx": _pkg_version("python-docx"),
+        "mammoth": _pkg_version("mammoth"),
+        "requests": _pkg_version("requests"),
+    }
+
diff --git a/docs/deployment.md b/docs/deployment.md
index 0c399e3..4fb2b61 100644
--- a/docs/deployment.md
+++ b/docs/deployment.md
@@ -7,6 +7,7 @@
 - **Automated Deployment Script:** `bash ~/app/scripts/deploy-app.sh`
 - **Manual Maintenance Toggle:** `bash ~/app/scripts/maintenance-mode.sh {on|off|status}`
 - **Complete Strategy & Troubleshooting:** The full deployment strategy is maintained in the repository documentation for operators working from the source tree.
+- **Shared-core retrofit rollout checklist:** `docs/shared-core-retrofit.md` (server deployment readiness and smoke tests)
 
 ---
 
diff --git a/docs/shared-core-retrofit.md b/docs/shared-core-retrofit.md
new file mode 100644
index 0000000..7ffa1fb
--- /dev/null
+++ b/docs/shared-core-retrofit.md
@@ -0,0 +1,94 @@
+# Shared Core Retrofit Plan and Status
+
+## Goal
+
+Retrofit GLOW so desktop CLI, desktop GUI, and web routes use a shared internal service layer for document audit/fix/convert dispatch, preparing extraction into a standalone cross-product core.
+
+## Implemented in this retrofit
+
+## 1) Shared package introduced
+
+- Added `desktop/src/acb_large_print_core/` with:
+  - `services.py`: canonical extension-based dispatch for audit/fix/convert
+  - `versions.py`: normalized component-version manifest for diagnostics
+  - `__init__.py`: exported core APIs
+
+## 2) Desktop CLI wired to shared services
+
+- `acb_large_print.cli` now routes audit/fix/convert dispatch through:
+  - `acb_large_print_core.services.audit_by_extension`
+  - `acb_large_print_core.services.fix_by_extension`
+  - `acb_large_print_core.services.convert_to_markdown`
+
+## 3) Web routes/tasks wired to shared services
+
+- `routes/audit.py` now uses shared `audit_by_extension`.
+- `routes/fix.py` now uses shared `audit_by_extension` + `fix_by_extension`.
+- `routes/convert.py`, `upload.py`, `routes/speech.py`, and `magic_features.py` now use shared conversion entry points.
+- `tasks/convert_tasks.py` now uses shared audit/markdown conversion dispatch.
+- `chat_handler.py` now routes live audit via shared `audit_by_extension`.
+
+## 4) Version-provenance plumbing
+
+- Added `acb_large_print_core.versions.get_component_versions()`.
+- Web template context now includes `component_versions` for About/support surfaces.
+- Desktop package `acb_large_print.__version__` now resolves from repository `VERSION` file.
+
+## 5) MCP server wiring
+
+- `mcp_server/glow_mcp_utils.py` now dispatches docx/markdown audit, docx fix, and markdown conversion via `acb_large_print_core.services`.
+
+## Contract surface (internal v1)
+
+Use these as canonical internal service contracts:
+
+- `audit_by_extension(path, **policy_overrides) -> AuditResult`
+- `fix_by_extension(path, output_path=None, **fix_options) -> tuple`
+- `convert_to_markdown(path, output_path=None) -> tuple[Path, str]`
+- `get_component_versions() -> dict[str, str]`
+
+## Next extraction step
+
+When ready to externalize for QUILL integration:
+
+1. Move `acb_large_print_core` to its own package/repo.
+2. Preserve function signatures and return types.
+3. Keep thin adapter modules in GLOW (`acb_large_print`) to prevent breaking consumers.
+4. Add semver and compatibility matrix (`glow_min`, `quill_min`) to version manifest.
+
+## Server deployment readiness checklist (for PR review and rollout)
+
+Use this checklist before promoting `feature/shared-core-retrofit` to production:
+
+1. **Branch + PR readiness**
+   - Confirm PR #81 is approved and mergeable.
+   - Confirm deployment target references the merged commit SHA (not a stale image/revision).
+
+2. **Build/runtime alignment**
+   - Rebuild application image/environment from the merged commit.
+   - Verify runtime imports resolve to repository source used in deployment (avoid stale site-packages path drift).
+
+3. **Automated verification gates**
+   - Desktop targeted checks:
+     - `python -m pytest tests/test_conversion_format_support.py tests/test_pdf_table_extraction.py -q`
+   - Web targeted checks:
+     - `python -m pytest tests/test_upload.py tests/test_fix_routes.py -q`
+   - Version surface checks (from repo root with source paths):
+     - `PYTHONPATH=desktop/src;web/src python -m pytest web/tests/test_app.py -k release_from_version_file -q`
+
+4. **Runtime smoke tests (staging first)**
+   - Audit flow works for `.docx`, `.md`, `.pdf`.
+   - Fix flow works for `.docx`; advisory/manual paths remain correct for unsupported auto-fix formats.
+   - Convert flow produces markdown via shared service dispatch.
+   - Chat audit summary still functions.
+   - MCP utility endpoints still run docx/markdown audit/fix/convert paths.
+
+5. **Observability + metadata**
+   - About page renders component versions (`release`, `markitdown`, `pymupdf`, core package).
+   - Logs show no import-dispatch errors for `acb_large_print_core`.
+
+6. **Rollout strategy**
+   - Deploy to staging.
+   - Run smoke checks above.
+   - Promote to production.
+   - Monitor first 24h for audit/fix/convert error-rate regressions.
diff --git a/mcp_server/glow_mcp_utils.py b/mcp_server/glow_mcp_utils.py
index cbb63e4..5748bbb 100644
--- a/mcp_server/glow_mcp_utils.py
+++ b/mcp_server/glow_mcp_utils.py
@@ -6,20 +6,30 @@
 """
 from pathlib import Path
 import tempfile
-import shutil
 import sys
 
-
-from acb_large_print.auditor import audit_document
-from acb_large_print.md_auditor import audit_markdown
-from acb_large_print.fixer import fix_document
-from acb_large_print.converter import convert_to_markdown
 from acb_large_print.pandoc_converter import convert_to_html, convert_to_docx
 from acb_large_print.reporter import generate_json_report, generate_text_report, generate_html_report
 
 SUPPORTED_FORMATS = {"markdown", "md", "docx", "html"}
 
 
+def _resolve_core_services():
+    """Return (audit_by_extension, convert_to_markdown, fix_by_extension).
+
+    Import lazily so module import does not fail in environments where shared
+    core packages are not installed but endpoints that need them are not used.
+    """
+    try:
+        from quill_glow_core import audit_by_extension, convert_to_markdown, fix_by_extension
+
+        return audit_by_extension, convert_to_markdown, fix_by_extension
+    except Exception as exc:
+        raise RuntimeError(
+            "Shared core services unavailable. Install quill-glow-core for MCP operations."
+        ) from exc
+
+
 def run_page_flow_extract(source_url: str, *, max_pages: int = 5, follow_pagination: bool = True):
     """Extract readable article text from a web URL using PageFlow logic.
 
@@ -66,24 +76,25 @@ def run_page_flow_extract(source_url: str, *, max_pages: int = 5, follow_paginat
 
 def run_audit(file_path: Path, fmt: str):
     """Dispatch to the correct audit function based on format."""
+    audit_by_extension, _convert_to_markdown, _fix_by_extension = _resolve_core_services()
     fmt = fmt.lower()
-    if fmt in ("markdown", "md"):
-        return audit_markdown(file_path)
-    if fmt == "docx":
-        return audit_document(file_path)
+    if fmt in ("markdown", "md", "docx"):
+        return audit_by_extension(file_path)
     raise ValueError(f"Unsupported format for audit: {fmt}")
 
 
 def run_fix(file_path: Path, fmt: str, output_path: Path = None):
     """Dispatch to the correct fix function based on format."""
+    _audit_by_extension, _convert_to_markdown, fix_by_extension = _resolve_core_services()
     fmt = fmt.lower()
     if fmt == "docx":
-        return fix_document(file_path, output_path)
+        return fix_by_extension(file_path, output_path=output_path)
     raise ValueError(f"Unsupported format for fix: {fmt}")
 
 
 def run_convert(file_path: Path, from_fmt: str, to_fmt: str, output_path: Path = None):
     """Dispatch to the correct convert function based on formats."""
+    _audit_by_extension, convert_to_markdown, _fix_by_extension = _resolve_core_services()
     from_fmt = from_fmt.lower()
     to_fmt = to_fmt.lower()
     if from_fmt == "docx" and to_fmt in ("markdown", "md"):
diff --git a/mcp_server/main.py b/mcp_server/main.py
index 545fe16..1b91c10 100644
--- a/mcp_server/main.py
+++ b/mcp_server/main.py
@@ -18,7 +18,6 @@
 from fastapi import FastAPI, UploadFile, File, Form
 from fastapi.responses import JSONResponse
 from fastapi.middleware.cors import CORSMiddleware
-from typing import Optional
 import tempfile
 from pathlib import Path
 try:
@@ -26,6 +25,27 @@
 except ImportError:  # Support direct module execution in tests.
     from glow_mcp_utils import run_audit, run_report, run_page_flow_extract
 
+try:
+    from quill_glow_core import (
+        configure_default_services as _configure_shared_core_default,
+        get_startup_telemetry_dict as _get_shared_core_startup_telemetry,
+    )
+except Exception:
+    _configure_shared_core_default = None
+
+    def _get_shared_core_startup_telemetry() -> dict:
+        return {
+            "backend": "unknown",
+            "configured_by": "unknown",
+            "auto_selected": None,
+        }
+
+if _configure_shared_core_default is not None:
+    try:
+        _configure_shared_core_default()
+    except Exception:
+        pass
+
 app = FastAPI(title="GLOW MCP Server", description="Accessibility audit/fix/convert/report API for agent integration.", version="7.2.0")
 
 app.add_middleware(
@@ -39,7 +59,10 @@
 @app.get("/health")
 def health():
     """Health check endpoint."""
-    return {"status": "ok"}
+    return {
+        "status": "ok",
+        "shared_core": _get_shared_core_startup_telemetry(),
+    }
 
 
 @app.post("/audit")
diff --git a/mcp_server/requirements.txt b/mcp_server/requirements.txt
index dccfa79..987e07c 100644
--- a/mcp_server/requirements.txt
+++ b/mcp_server/requirements.txt
@@ -1,4 +1,5 @@
 # requirements.txt for GLOW MCP Server
+quill-glow-core>=0.1.0
 fastapi==0.110.2
 uvicorn==0.29.0
 python-multipart==0.0.28
diff --git a/web/pyproject.toml b/web/pyproject.toml
index 825968b..396dc8f 100644
--- a/web/pyproject.toml
+++ b/web/pyproject.toml
@@ -22,6 +22,7 @@ classifiers = [
     "Topic :: Adaptive Technologies",
 ]
 dependencies = [
+    "quill-glow-core>=0.1.0",
     "flask>=3.1",
     "flask-wtf>=1.2",
     "flask-limiter>=4.0",
diff --git a/web/requirements.txt b/web/requirements.txt
index 01ba60e..636f718 100644
--- a/web/requirements.txt
+++ b/web/requirements.txt
@@ -1,3 +1,4 @@
+quill-glow-core>=0.1.0
 flask>=3.1
 flask-wtf>=1.2
 flask-limiter>=3.5
diff --git a/web/src/acb_large_print_web/app.py b/web/src/acb_large_print_web/app.py
index 1ebeadf..5fbb66a 100644
--- a/web/src/acb_large_print_web/app.py
+++ b/web/src/acb_large_print_web/app.py
@@ -17,6 +17,15 @@
 
 from .rules import get_help_urls_map, get_rules_by_category, get_rules_by_severity
 
+try:
+    from quill_glow_core import (
+        configure_default_services as _configure_shared_core_default,
+        get_startup_telemetry_dict as _get_shared_core_startup_telemetry,
+    )
+except Exception:
+    _configure_shared_core_default = None
+    _get_shared_core_startup_telemetry = None
+
 csrf = CSRFProtect()
 limiter = Limiter(
     key_func=get_remote_address,
@@ -60,6 +69,13 @@ def create_app(config: dict | None = None) -> Flask:
     if config:
         app.config.update(config)
 
+    # Initialize shared-core wiring once at process startup when available.
+    if _configure_shared_core_default is not None:
+        try:
+            _configure_shared_core_default()
+        except Exception:
+            app.logger.exception("Failed to initialize quill-glow-core shared services")
+
     # Extensions
     csrf.init_app(app)
     limiter.init_app(app)
@@ -151,6 +167,7 @@ def _log_request(response):
     @app.context_processor
     def inject_rules():
         from flask import g as _g
+        from quill_glow_core import get_component_versions as _get_component_versions
         from .ai_features import get_all_flags as _get_ai_flags
         from .branding import get_branding_context as _get_branding_context
         from .version import get_version as _get_release_version
@@ -173,6 +190,7 @@ def inject_rules():
             "web_version": web_ver,
             "desktop_version": desktop_ver,
             "release_version": release_ver,
+            "component_versions": _get_component_versions(),
             "csp_nonce": getattr(_g, "csp_nonce", ""),
         }
         # Inject AI flags (from ai_features)
@@ -801,6 +819,15 @@ def _build_health_payload() -> tuple[dict, bool]:
             "timestamp_utc": datetime.now(UTC).isoformat(),
             "duration_ms": _hduration_ms,
         }
+        if _get_shared_core_startup_telemetry is not None:
+            try:
+                payload["shared_core"] = _get_shared_core_startup_telemetry()
+            except Exception:
+                payload["shared_core"] = {
+                    "backend": "unknown",
+                    "configured_by": "unknown",
+                    "auto_selected": None,
+                }
         return payload, all_ok
 
     # Health check
diff --git a/web/src/acb_large_print_web/chat_handler.py b/web/src/acb_large_print_web/chat_handler.py
index e1fe321..72173a6 100644
--- a/web/src/acb_large_print_web/chat_handler.py
+++ b/web/src/acb_large_print_web/chat_handler.py
@@ -488,9 +488,9 @@ def run_accessibility_audit(self) -> str:
         if self.context.doc_path is None:
             return self._heuristic_compliance_summary()
         try:
-            from acb_large_print.auditor import audit_document
+            from quill_glow_core import audit_by_extension
 
-            result = audit_document(str(self.context.doc_path))
+            result = audit_by_extension(str(self.context.doc_path))
             self.context._audit_cache = {
                 "findings": [
                     {
diff --git a/web/src/acb_large_print_web/magic_features.py b/web/src/acb_large_print_web/magic_features.py
index a3521f0..3a41dac 100644
--- a/web/src/acb_large_print_web/magic_features.py
+++ b/web/src/acb_large_print_web/magic_features.py
@@ -320,7 +320,7 @@ def extract_text_for_compare(path: Path) -> str:
         return path.read_text(encoding="utf-8", errors="replace")
 
     # Use existing MarkItDown extractor for binary docs.
-    from acb_large_print.converter import convert_to_markdown
+    from quill_glow_core import convert_to_markdown
 
     out_path = path.with_suffix(path.suffix + ".cmp.md")
     md_path, _ = convert_to_markdown(path, output_path=out_path)
diff --git a/web/src/acb_large_print_web/routes/audit.py b/web/src/acb_large_print_web/routes/audit.py
index 7f67af8..d889347 100644
--- a/web/src/acb_large_print_web/routes/audit.py
+++ b/web/src/acb_large_print_web/routes/audit.py
@@ -264,31 +264,9 @@ def _compute_audit_diff(
 
 def _audit_by_extension(saved_path: Path):
     """Dispatch to the correct auditor based on file extension."""
-    ext = saved_path.suffix.lower()
-    if ext == ".xlsx":
-        from acb_large_print.xlsx_auditor import audit_workbook
+    from quill_glow_core import audit_by_extension
 
-        return audit_workbook(saved_path)
-    elif ext == ".pptx":
-        from acb_large_print.pptx_auditor import audit_presentation
-
-        return audit_presentation(saved_path)
-    elif ext == ".md":
-        from acb_large_print.md_auditor import audit_markdown
-
-        return audit_markdown(saved_path)
-    elif ext == ".pdf":
-        from acb_large_print.pdf_auditor import audit_pdf
-
-        return audit_pdf(saved_path)
-    elif ext == ".epub":
-        from acb_large_print.epub_auditor import audit_epub
-
-        return audit_epub(saved_path)
-    else:
-        from acb_large_print.auditor import audit_document
-
-        return audit_document(saved_path)
+    return audit_by_extension(saved_path)
 
 
 def _format_from_path(saved_path: Path) -> str:
diff --git a/web/src/acb_large_print_web/routes/convert.py b/web/src/acb_large_print_web/routes/convert.py
index 9044a85..da16779 100644
--- a/web/src/acb_large_print_web/routes/convert.py
+++ b/web/src/acb_large_print_web/routes/convert.py
@@ -28,11 +28,9 @@
 
 from werkzeug.utils import secure_filename as _secure_filename
 
-from acb_large_print.converter import (
-    CONVERTIBLE_EXTENSIONS,
-    MARKITDOWN_AUDIO_EXTENSIONS,
-    convert_to_markdown,
-)
+from acb_large_print.converter import CONVERTIBLE_EXTENSIONS, MARKITDOWN_AUDIO_EXTENSIONS
+
+from quill_glow_core import convert_to_markdown
 from acb_large_print.wcag_language import (
     analyze_text_for_wcag_language,
 )
diff --git a/web/src/acb_large_print_web/routes/fix.py b/web/src/acb_large_print_web/routes/fix.py
index 2638ff2..7986801 100644
--- a/web/src/acb_large_print_web/routes/fix.py
+++ b/web/src/acb_large_print_web/routes/fix.py
@@ -126,107 +126,36 @@ def _fix_by_extension(
 
     Returns (output_path, total_fixes, fix_records, post_audit, warnings).
     """
-    ext = saved_path.suffix.lower()
-    if ext == ".xlsx":
-        from acb_large_print.xlsx_auditor import audit_workbook
-
-        post_audit = audit_workbook(saved_path)
-        return (
-            saved_path,
-            0,
-            [],
-            post_audit,
-            [
-                "Excel workbooks cannot be auto-fixed yet. "
-                "Review the audit findings and fix them manually in Excel."
-            ],
-        )
-    elif ext == ".pptx":
-        from acb_large_print.pptx_auditor import audit_presentation
+    from quill_glow_core import fix_by_extension
 
-        post_audit = audit_presentation(saved_path)
-        return (
-            saved_path,
-            0,
-            [],
-            post_audit,
-            [
-                "PowerPoint presentations cannot be auto-fixed yet. "
-                "Review the audit findings and fix them manually in PowerPoint."
-            ],
-        )
-    elif ext == ".md":
-        from acb_large_print.md_auditor import audit_markdown
-
-        post_audit = audit_markdown(saved_path)
-        return (
-            saved_path,
-            0,
-            [],
-            post_audit,
-            [
-                "Markdown auto-fix is coming soon. "
-                "Review the audit findings and fix them in your text editor."
-            ],
-        )
-    elif ext == ".pdf":
-        from acb_large_print.pdf_auditor import audit_pdf
-
-        post_audit = audit_pdf(saved_path)
-        return (
-            saved_path,
-            0,
-            [],
-            post_audit,
-            [
-                "PDF files cannot be auto-fixed. "
-                "Use Adobe Acrobat Pro or re-export from the source application."
-            ],
-        )
-    elif ext == ".epub":
-        from acb_large_print.epub_auditor import audit_epub
-
-        post_audit = audit_epub(saved_path)
-        return (
-            saved_path,
-            0,
-            [],
-            post_audit,
-            [
-                "ePub files cannot be auto-fixed yet. "
-                "Review the audit findings and fix them in your ePub editor."
-            ],
-        )
-    else:
-        from acb_large_print.fixer import fix_document
+    ai_provider = None
+    if detect_headings and use_ai:
+        try:
+            from acb_large_print.ai_provider import get_provider
 
-        ai_provider = None
-        if detect_headings and use_ai:
-            try:
-                from acb_large_print.ai_provider import get_provider
+            ai_provider = get_provider()
+        except Exception:
+            pass  # Fall back to heuristic-only
 
-                ai_provider = get_provider()
-            except Exception:
-                pass  # Fall back to heuristic-only
-        result = fix_document(
-            saved_path,
-            output_path,
-            bound=bound,
-            list_indent_in=list_indent_in,
-            list_hanging_in=list_hanging_in,
-            list_level_indents=list_level_indents,
-            para_indent_in=para_indent_in,
-            first_line_indent_in=first_line_indent_in,
-            preserve_heading_alignment=preserve_heading_alignment,
-            detect_headings=detect_headings,
-            ai_provider=ai_provider,
-            heading_threshold=heading_threshold,
-            confirmed_headings=confirmed_headings,
-            heading_accuracy_level=heading_accuracy,
-            style_size_overrides=style_size_overrides,
-        )
-        # Tag whether AI was actually used (ai_provider set and invoked)
-        return result[:5] + ({"ai_used": ai_provider is not None},) if len(result) == 5 else result
+    result = fix_by_extension(
+        saved_path,
+        output_path,
+        bound=bound,
+        list_indent_in=list_indent_in,
+        list_hanging_in=list_hanging_in,
+        list_level_indents=list_level_indents,
+        para_indent_in=para_indent_in,
+        first_line_indent_in=first_line_indent_in,
+        preserve_heading_alignment=preserve_heading_alignment,
+        detect_headings=detect_headings,
+        ai_provider=ai_provider,
+        heading_threshold=heading_threshold,
+        confirmed_headings=confirmed_headings,
+        heading_accuracy_level=heading_accuracy,
+        style_size_overrides=style_size_overrides,
+    )
+    # Tag whether AI was actually used (ai_provider set and invoked)
+    return result[:5] + ({"ai_used": ai_provider is not None},) if len(result) == 5 else result
 
 
 def _audit_by_extension(
@@ -239,38 +168,16 @@ def _audit_by_extension(
     style_size_overrides: dict[str, float] | None = None,
 ):
     """Dispatch to the correct auditor based on file extension."""
-    ext = saved_path.suffix.lower()
-    if ext == ".xlsx":
-        from acb_large_print.xlsx_auditor import audit_workbook
-
-        return audit_workbook(saved_path)
-    elif ext == ".pptx":
-        from acb_large_print.pptx_auditor import audit_presentation
-
-        return audit_presentation(saved_path)
-    elif ext == ".md":
-        from acb_large_print.md_auditor import audit_markdown
+    from quill_glow_core import audit_by_extension
 
-        return audit_markdown(saved_path)
-    elif ext == ".pdf":
-        from acb_large_print.pdf_auditor import audit_pdf
-
-        return audit_pdf(saved_path)
-    elif ext == ".epub":
-        from acb_large_print.epub_auditor import audit_epub
-
-        return audit_epub(saved_path)
-    else:
-        from acb_large_print.auditor import audit_document
-
-        return audit_document(
-            saved_path,
-            list_indent_in=list_indent_in,
-            list_level_indents=list_level_indents,
-            para_indent_in=para_indent_in,
-            first_line_indent_in=first_line_indent_in,
-            style_size_overrides=style_size_overrides,
-        )
+    return audit_by_extension(
+        saved_path,
+        list_indent_in=list_indent_in,
+        list_level_indents=list_level_indents,
+        para_indent_in=para_indent_in,
+        first_line_indent_in=first_line_indent_in,
+        style_size_overrides=style_size_overrides,
+    )
 
 
 def _format_from_path(saved_path: Path) -> str:
diff --git a/web/src/acb_large_print_web/routes/speech.py b/web/src/acb_large_print_web/routes/speech.py
index 5194994..9cefe87 100644
--- a/web/src/acb_large_print_web/routes/speech.py
+++ b/web/src/acb_large_print_web/routes/speech.py
@@ -30,7 +30,7 @@
     url_for,
 )
 
-from acb_large_print.converter import convert_to_markdown
+from quill_glow_core import convert_to_markdown
 from acb_large_print.pandoc_converter import pandoc_available
 
 from ..app import limiter
diff --git a/web/src/acb_large_print_web/tasks/convert_tasks.py b/web/src/acb_large_print_web/tasks/convert_tasks.py
index cd242b4..b3f42d7 100644
--- a/web/src/acb_large_print_web/tasks/convert_tasks.py
+++ b/web/src/acb_large_print_web/tasks/convert_tasks.py
@@ -376,13 +376,15 @@ def _dispatch_conversion(
     options: dict[str, Any],
 ) -> str:
     """Run the conversion and return the absolute path of the result file."""
-    from ..upload import get_temp_dir
-    from acb_large_print.converter import CONVERTIBLE_EXTENSIONS, convert_to_markdown
+    from acb_large_print.converter import CONVERTIBLE_EXTENSIONS
     from acb_large_print.pandoc_converter import (
-        PANDOC_INPUT_EXTENSIONS,
         LIBREOFFICE_CONVERSIONS,
+        PANDOC_INPUT_EXTENSIONS,
         preconvert_via_libreoffice,
     )
+    from quill_glow_core import convert_to_markdown
+
+    from ..upload import get_temp_dir
 
     _progress(job_id, 10, "Locating source file…")
     temp_dir = get_temp_dir(upload_token)
@@ -460,8 +462,7 @@ def _dispatch_conversion(
 
 
 def _run_pipeline(job_id, source, out_dir, options):
-    from acb_large_print.pipeline_converter import convert_with_pipeline
-    from acb_large_print.pipeline_converter import get_available_conversions
+    from acb_large_print.pipeline_converter import convert_with_pipeline, get_available_conversions
     _progress(job_id, 30, "Sending to DAISY Pipeline…")
     conversion_key = options.get("pipeline_conversion", "")
     available = get_available_conversions()
@@ -480,7 +481,7 @@ def _run_pipeline(job_id, source, out_dir, options):
 
 
 def _run_to_markdown(job_id, source, out_dir, options):
-    from acb_large_print.converter import convert_to_markdown
+    from quill_glow_core import convert_to_markdown
     _progress(job_id, 40, "Extracting content to Markdown…")
     dest = out_dir / (source.stem + ".md")
     output_path, _ = convert_to_markdown(source, output_path=dest)
@@ -489,6 +490,7 @@ def _run_to_markdown(job_id, source, out_dir, options):
 
 def _run_to_html(job_id, source, out_dir, options):
     import re
+
     from acb_large_print.pandoc_converter import convert_to_html
     _progress(job_id, 40, "Converting to HTML…")
     css_path = None if options.get("acb_format", True) else Path("__no_acb_css__")
@@ -692,15 +694,15 @@ def _run_speech(
     output_format: str,
 ) -> str:
     """Run full-document speech synthesis and return the output file path."""
-    from ..upload import get_temp_dir
     from acb_large_print_web.speech import (
         SpeechError,
         normalize_document_text,
         synthesize_document_text,
         wav_bytes_to_mp3,
-        wav_duration_seconds,
     )
 
+    from ..upload import get_temp_dir
+
     _progress(job_id, 10, "Locating extracted document text…")
 
     # load pre-extracted text written by /speech/prepare
@@ -727,8 +729,8 @@ def _run_speech(
 
     # Optionally apply pronunciation dictionary if enabled
     try:
-        from acb_large_print_web.magic_features import apply_pronunciation_dictionary
         from acb_large_print_web import feature_flags as _ff
+        from acb_large_print_web.magic_features import apply_pronunciation_dictionary
         if _ff.get_all_flags().get("GLOW_ENABLE_SPEECH_PRONUNCIATION_DICTIONARY", True):
             text = apply_pronunciation_dictionary(text)
     except Exception:
@@ -840,10 +842,12 @@ def run_export_job(
     token: str,
     options: dict[str, Any],
 ) -> dict[str, Any]:
-    from ..upload import get_temp_dir
-    from acb_large_print.exporter import export_cms_fragment, export_standalone_html
     import zipfile
 
+    from acb_large_print.exporter import export_cms_fragment, export_standalone_html
+
+    from ..upload import get_temp_dir
+
     status = read_status(job_id)
     max_attempts = max(1, int(status.get("max_attempts", 1)))
     start_attempt = max(1, int(status.get("attempt", 0)) + 1)
@@ -896,28 +900,10 @@ def run_audit_job(
     input_filename: str,
     options: dict[str, Any],
 ) -> dict[str, Any]:
-    from ..upload import get_temp_dir
-    from acb_large_print.auditor import audit_document
-    from acb_large_print.md_auditor import audit_markdown
-    from acb_large_print.pptx_auditor import audit_presentation
-    from acb_large_print.pdf_auditor import audit_pdf
-    from acb_large_print.epub_auditor import audit_epub
-    from acb_large_print.xlsx_auditor import audit_workbook
     from acb_large_print.constants import AUDIT_RULES
+    from quill_glow_core import audit_by_extension
 
-    def _audit_by_ext(path: Path):
-        ext = path.suffix.lower()
-        if ext == ".xlsx":
-            return audit_workbook(path)
-        if ext == ".pptx":
-            return audit_presentation(path)
-        if ext == ".md":
-            return audit_markdown(path)
-        if ext == ".pdf":
-            return audit_pdf(path)
-        if ext == ".epub":
-            return audit_epub(path)
-        return audit_document(path)
+    from ..upload import get_temp_dir
 
     status = read_status(job_id)
     max_attempts = max(1, int(status.get("max_attempts", 1)))
@@ -934,7 +920,7 @@ def _audit_by_ext(path: Path):
             if not src.exists():
                 raise FileNotFoundError(f"Source file not found: {src}")
             write_status(job_id, state="PROGRESS", progress=25, message="Running audit…", retryable=True)
-            result = _audit_by_ext(src)
+            result = audit_by_extension(src)
             findings = _serialize_findings(getattr(result, "findings", []))
             summary = {
                 "filename": input_filename,
@@ -969,8 +955,9 @@ def run_fix_job(
     input_filename: str,
     options: dict[str, Any],
 ) -> dict[str, Any]:
+    from acb_large_print_web.routes.fix import _fix_by_extension
+
     from ..upload import get_temp_dir
-    from acb_large_print_web.routes.fix import _fix_by_extension, _audit_by_extension
 
     status = read_status(job_id)
     max_attempts = max(1, int(status.get("max_attempts", 1)))
@@ -1046,14 +1033,15 @@ def run_speech_prepare_job(
     input_filename: str,
     speed: float,
 ) -> dict[str, Any]:
-    from ..upload import get_temp_dir
+    from acb_large_print_web.routes.speech import _DOC_EXTRACT_NAME, _DOC_RENDERED_NAME, _extract_document_text
     from acb_large_print_web.speech import (
-        normalize_document_text,
         estimate_audio_seconds_from_text,
         estimate_processing_seconds_from_text,
         first_sentences,
+        normalize_document_text,
     )
-    from acb_large_print_web.routes.speech import _extract_document_text, _DOC_EXTRACT_NAME, _DOC_RENDERED_NAME
+
+    from ..upload import get_temp_dir
 
     status = read_status(job_id)
     max_attempts = max(1, int(status.get("max_attempts", 1)))
diff --git a/web/src/acb_large_print_web/templates/about.html b/web/src/acb_large_print_web/templates/about.html
index 3f748a7..9b008a3 100644
--- a/web/src/acb_large_print_web/templates/about.html
+++ b/web/src/acb_large_print_web/templates/about.html
@@ -435,6 +435,15 @@ <h2 id="project-heading">Project Information</h2>
     <dt>Web application version</dt>
     <dd>{{ web_version }}</dd>
 
+    <dt>MarkItDown version</dt>
+    <dd>{{ component_versions.markitdown if component_versions else "unknown" }}</dd>
+
+    <dt>PyMuPDF version</dt>
+    <dd>{{ component_versions.pymupdf if component_versions else "unknown" }}</dd>
+
+    <dt>Shared service core package</dt>
+    <dd>{{ component_versions.desktop_package if component_versions else "unknown" }}</dd>
+
     <dt>Python version</dt>
     <dd>3.13+</dd>
   </dl>
diff --git a/web/src/acb_large_print_web/upload.py b/web/src/acb_large_print_web/upload.py
index 89810ae..83f223a 100644
--- a/web/src/acb_large_print_web/upload.py
+++ b/web/src/acb_large_print_web/upload.py
@@ -9,8 +9,10 @@
 from datetime import datetime
 from pathlib import Path
 
-from acb_large_print.converter import CONVERTIBLE_EXTENSIONS as MARKITDOWN_CONVERTIBLE_EXTENSIONS
-from acb_large_print.converter import MARKITDOWN_AUDIO_EXTENSIONS
+from acb_large_print.converter import (
+    CONVERTIBLE_EXTENSIONS as MARKITDOWN_CONVERTIBLE_EXTENSIONS,
+    MARKITDOWN_AUDIO_EXTENSIONS,
+)
 
 from werkzeug.datastructures import FileStorage
 from werkzeug.utils import secure_filename