diff --git a/.github/workflows/markdown-parser.yml b/.github/workflows/markdown-parser.yml
new file mode 100644
index 00000000..e5a3f3ad
--- /dev/null
+++ b/.github/workflows/markdown-parser.yml
@@ -0,0 +1,112 @@
+name: Markdown parser tests
+
+on:
+  pull_request:
+  push:
+    branches: ["master"]
+  workflow_dispatch:
+
+jobs:
+  test:
+    runs-on: ubuntu-20.04
+
+    steps:
+      - name: Checkout benchmarks project
+        uses: actions/checkout@v2
+        with:
+          path: benchmarks
+
+      # - name: Install node
+      #   uses: actions/setup-node@v2
+      #   with:
+      #     node-version: "14.x"
+
+      - name: Install Python
+        uses: actions/setup-python@v2
+        with:
+          python-version: 3.8
+
+      - name: Cache pip on Linux
+        uses: actions/cache@v1
+        with:
+          path: ~/.cache/pip
+          key: ${{ runner.os }}-pip-3.8-${{ hashFiles('**/requirements.txt', 'setup.cfg') }}
+          restore-keys: |
+            ${{ runner.os }}-pip-3.8
+
+      # - name: Get yarn cache directory path
+      #   id: yarn-cache-dir-path
+      #   run: echo "::set-output name=dir::$(yarn cache dir)"
+      # - name: Cache yarn
+      #   uses: actions/cache@v1
+      #   id: yarn-cache # use this to check for `cache-hit` (`steps.yarn-cache.outputs.cache-hit != 'true'`)
+      #   with:
+      #     path: ${{ steps.yarn-cache-dir-path.outputs.dir }}
+      #     key: ${{ runner.os }}-yarn-${{ hashFiles('**/yarn.lock') }}
+      #     restore-keys: |
+      #       ${{ runner.os }}-yarn-
+
+      - name: Install dependencies
+        working-directory: benchmarks/markdown-parser
+        run: |
+          set -ex
+          python -m pip install -r requirements.txt
+          python -m playwright install chromium
+
+      # - name: Checkout JupyterLab
+      #   uses: actions/checkout@v2
+      #   with:
+      #     repository: jupyterlab/jupyterlab
+      #     ref: master
+      #     path: reference
+
+      # - name: Install dependencies
+      #   run: |
+      #     set -ex
+      #     echo "OLD_REF_SHA=$(git log -n1 --format='%H')" >> $GITHUB_ENV
+      #     bash ./scripts/ci_install.sh
+      #     # Build dev mode
+      #     jlpm run build
+      #   working-directory: reference
+
+      - name: Launch JupyterLab
+        shell: bash
+        run: |
+          set -ex
+          python -m jupyterlab --config jupyter_lab_config.py 2>&1 > /tmp/jupyterlab_server.log &
+        working-directory: benchmarks/markdown-parser
+
+      - name: Wait for JupyterLab
+        uses: ifaxity/wait-on-action@v1
+        with:
+          resource: http-get://localhost:9999/lab
+          timeout: 360000
+
+      - name: Tests
+        working-directory: benchmarks/markdown-parser
+        run: |
+          set -ex
+          mkdir -p reports
+          # For now limit ot JupyterLab vs GFM
+          pytest -rap -vv --rootdir "$PWD" --base-url "http://localhost:9999" --report-dir "$PWD/reports" tests/test_jupyterlab.py
+
+      - name: Upload pytest reports
+        if: always()
+        uses: actions/upload-artifact@v2
+        with:
+          name: markdown-parser-report-${{ github.run_number }}
+          path: |
+            benchmarks/markdown-parser/reports
+
+      - name: Kill the server
+        if: always()
+        shell: bash
+        run: |
+          kill -s SIGKILL $(pgrep jupyter-lab)
+
+      - name: Print JupyterLab logs
+        if: always()
+        run: |
+          echo "::group::Server log"
+          cat /tmp/jupyterlab_server.log
+          echo "::endgroup::"
diff --git a/markdown-parser/README.md b/markdown-parser/README.md
new file mode 100644
index 00000000..d671d9c9
--- /dev/null
+++ b/markdown-parser/README.md
@@ -0,0 +1,8 @@
+# Markdown parser tests
+
+This folder contains tests to validate markdown parsers.
+
+The tests are written using [pytest](https://docs.pytest.org). To test web frontend parsers, [playwright](https://playwright.dev/python/docs/intro) is used to evaluate the markdown conversion to HTML.
+
+> The test database is downloaded from [GitHub flavored Commonmark](https://github.com/github/cmark-gfm).
+
diff --git a/markdown-parser/conftest.py b/markdown-parser/conftest.py
new file mode 100644
index 00000000..d699ea1c
--- /dev/null
+++ b/markdown-parser/conftest.py
@@ -0,0 +1,6 @@
+def pytest_addoption(parser):
+    """Add option to set the comparison reports"""
+    parser.addoption(
+        "--report-dir",
+        help="Directory in which the reports must be saved.",
+    )
diff --git a/markdown-parser/jupyter_lab_config.py b/markdown-parser/jupyter_lab_config.py
new file mode 100644
index 00000000..d7e46661
--- /dev/null
+++ b/markdown-parser/jupyter_lab_config.py
@@ -0,0 +1,21 @@
+import getpass
+from tempfile import mkdtemp
+
+# Test if we are running in a docker
+if getpass.getuser() == "jovyan":
+    c.ServerApp.ip = "0.0.0.0"
+
+c.ServerApp.port = 9999
+c.ServerApp.open_browser = False
+
+c.ServerApp.root_dir = mkdtemp(prefix='markdown-parser-lab-')
+
+c.ServerApp.token = ""
+c.ServerApp.password = ""
+c.ServerApp.disable_check_xsrf = True
+
+# c.LabApp.dev_mode = True
+# c.LabApp.extensions_in_dev_mode = True
+c.LabApp.expose_app_in_browser = True
+
+c.RetroApp.expose_app_in_browser = True
diff --git a/markdown-parser/requirements.txt b/markdown-parser/requirements.txt
new file mode 100644
index 00000000..a5e1e7c9
--- /dev/null
+++ b/markdown-parser/requirements.txt
@@ -0,0 +1,6 @@
+jupyterlab[test]
+nbconvert
+playwright
+pytest
+pytest-playwright
+requests
diff --git a/markdown-parser/tests/__init__.py b/markdown-parser/tests/__init__.py
new file mode 100644
index 00000000..e69de29b
diff --git a/markdown-parser/tests/conftest.py b/markdown-parser/tests/conftest.py
new file mode 100644
index 00000000..76ead114
--- /dev/null
+++ b/markdown-parser/tests/conftest.py
@@ -0,0 +1,117 @@
+import os
+import pathlib
+import pytest
+from playwright.sync_api import Error
+from slugify import slugify
+
+
+def _build_artifact_test_folder(pytestconfig, request, folder_or_file_name):
+    # Taken from pytest-playwright
+    output_dir = pytestconfig.getoption("--output")
+    return os.path.join(output_dir, slugify(request.node.nodeid), folder_or_file_name)
+
+
+@pytest.fixture(scope="session")
+def jupyterlab_page(browser, browser_context_args, pytestconfig, request):
+    # Merge pytest-playwright fixtures context and page at scope session and load the JupyterLab page
+    pages = []
+    context = browser.new_context(**browser_context_args)
+    context.on("page", lambda page: pages.append(page))
+
+    tracing_option = pytestconfig.getoption("--tracing")
+    capture_trace = tracing_option in ["on", "retain-on-failure"]
+    if capture_trace:
+        context.tracing.start(
+            name=slugify(request.node.nodeid),
+            screenshots=True,
+            snapshots=True,
+            sources=True,
+        )
+
+    page = context.new_page()
+    page.goto("/lab")
+    yield page
+
+    # If request.node is missing rep_call, then some error happened during execution
+    # that prevented teardown, but should still be counted as a failure
+    failed = request.node.rep_call.failed if hasattr(request.node, "rep_call") else True
+
+    if capture_trace:
+        retain_trace = tracing_option == "on" or (
+            failed and tracing_option == "retain-on-failure"
+        )
+        if retain_trace:
+            trace_path = _build_artifact_test_folder(pytestconfig, request, "trace.zip")
+            context.tracing.stop(path=trace_path)
+        else:
+            context.tracing.stop()
+
+    screenshot_option = pytestconfig.getoption("--screenshot")
+    capture_screenshot = screenshot_option == "on" or (
+        failed and screenshot_option == "only-on-failure"
+    )
+    if capture_screenshot:
+        for index, page in enumerate(pages):
+            human_readable_status = "failed" if failed else "finished"
+            screenshot_path = _build_artifact_test_folder(
+                pytestconfig, request, f"test-{human_readable_status}-{index+1}.png"
+            )
+            try:
+                page.screenshot(timeout=5000, path=screenshot_path)
+            except Error:
+                pass
+
+    context.close()
+
+    video_option = pytestconfig.getoption("--video")
+    preserve_video = video_option == "on" or (
+        failed and video_option == "retain-on-failure"
+    )
+    if preserve_video:
+        for page in pages:
+            video = page.video
+            if not video:
+                continue
+            try:
+                video_path = video.path()
+                file_name = os.path.basename(video_path)
+                video.save_as(
+                    path=_build_artifact_test_folder(pytestconfig, request, file_name)
+                )
+            except Error:
+                # Silent catch empty videos.
+                pass
+
+
+@pytest.fixture(scope="module")
+def md_report(request):
+    """Generate a comparison report for each test module.
+    
+    Each test must return a dictionary with the same keys. Each keys will be
+    a table header and each test will be a table row.
+    """
+    test_reports = []
+
+    yield test_reports
+
+    if len(test_reports[0]) > 0:
+        filename = pathlib.Path(request.config.getoption("report_dir")) / (
+            request.module.__name__.replace(".", "_") + "_report.md"
+        )
+
+        with filename.open("w") as f:
+            headers = test_reports[0]
+            f.writelines(
+                [
+                    f"# {request.module.__name__}\n",
+                    "\n",
+                    "| " + " | ".join(headers) + " |\n",
+                    "| " + " | ".join(["---"] * len(headers)) + " |\n",
+                ]
+            )
+            f.writelines(
+                map(
+                    lambda e: "| " + " | ".join(map(str, e.values())) + " |\n",
+                    test_reports,
+                )
+            )
diff --git a/markdown-parser/tests/test_jupyter.py b/markdown-parser/tests/test_jupyter.py
new file mode 100644
index 00000000..d72b5924
--- /dev/null
+++ b/markdown-parser/tests/test_jupyter.py
@@ -0,0 +1,32 @@
+import pytest
+from nbconvert.filters.markdown import markdown2html_mistune
+
+from .utils import commonmark_gfm_tests, get_jupyterlab_rendered_markdown
+
+
+@pytest.mark.parametrize("gfm", commonmark_gfm_tests())
+def test_nbconvert_jupyterlab(jupyterlab_page, md_report, gfm):
+
+    # Import normalize helper from github/cmark-gfm through its addition to sys.path in commonmark_gfm_tests
+    from normalize import normalize_html
+
+    given = gfm["markdown"]
+    test = normalize_html(markdown2html_mistune(given))
+    ref = normalize_html(jupyterlab_page.evaluate(get_jupyterlab_rendered_markdown, gfm["markdown"]))
+
+    success = True
+    try:
+        assert test == ref
+    except Exception as e:
+        success = False
+        raise e
+    finally:
+        md_report.append({
+            "id": gfm["example"],
+            "section": gfm["section"],
+            "failed": "" if success else "X",
+            "markdown": repr(given).replace("'", "`"),
+            "JupyterLab": repr(ref).replace("'", "`"),
+            "nbconvert - mistune": repr(test).replace("'", "`"),
+            "comments": ""
+        })
diff --git a/markdown-parser/tests/test_jupyterlab.py b/markdown-parser/tests/test_jupyterlab.py
new file mode 100644
index 00000000..7f94cbb4
--- /dev/null
+++ b/markdown-parser/tests/test_jupyterlab.py
@@ -0,0 +1,33 @@
+import pytest
+
+from .utils import commonmark_gfm_tests, get_jupyterlab_rendered_markdown
+
+
+@pytest.mark.parametrize("gfm", commonmark_gfm_tests())
+def test_gfm_jupyterlab_renderer(jupyterlab_page, md_report, gfm):
+
+    # Import normalize helper from github/cmark-gfm through its addition to sys.path in commonmark_gfm_tests
+    from normalize import normalize_html
+
+    given = gfm["markdown"]
+    test = normalize_html(jupyterlab_page.evaluate(get_jupyterlab_rendered_markdown, gfm["markdown"]))
+    ref = normalize_html(gfm["html"])
+
+    success = True
+    try:
+        assert test == ref
+    except Exception as e:
+        success = False
+        raise e
+    finally:
+        md_report.append(
+            {
+                "id": gfm["example"],
+                "section": gfm["section"],
+                "failed": "" if success else "X",
+                "markdown": repr(given).replace("'", "`"),
+                "commonmark-gfm": repr(ref).replace("'", "`"),
+                "JupyterLab": repr(test).replace("'", "`"),
+                "comments": "",
+            }
+        )
diff --git a/markdown-parser/tests/test_nbconvert.py b/markdown-parser/tests/test_nbconvert.py
new file mode 100644
index 00000000..205588a2
--- /dev/null
+++ b/markdown-parser/tests/test_nbconvert.py
@@ -0,0 +1,34 @@
+import pytest
+from nbconvert.filters.markdown import markdown2html_mistune
+
+from .utils import commonmark_gfm_tests
+
+
+@pytest.mark.parametrize("gfm", commonmark_gfm_tests())
+def test_gfm_nbconvert_markdown2html(md_report, gfm):
+
+    # Import normalize helper from github/cmark-gfm through its addition to sys.path in commonmark_gfm_tests
+    from normalize import normalize_html
+
+    given = gfm["markdown"]
+    test = normalize_html(markdown2html_mistune(given))
+    ref = normalize_html(gfm["html"])
+
+    success = True
+    try:
+        assert test == ref
+    except Exception as e:
+        success = False
+        raise e
+    finally:
+        md_report.append(
+            {
+                "id": gfm["example"],
+                "section": gfm["section"],
+                "failed": "" if success else "X",
+                "markdown": repr(given).replace("'", "`"),
+                "commonmark-gfm": repr(ref).replace("'", "`"),
+                "nbconvert - mistune": repr(test).replace("'", "`"),
+                "comments": "",
+            }
+        )
diff --git a/markdown-parser/tests/utils.py b/markdown-parser/tests/utils.py
new file mode 100644
index 00000000..96c9519a
--- /dev/null
+++ b/markdown-parser/tests/utils.py
@@ -0,0 +1,60 @@
+import functools
+from pathlib import Path
+import sys
+import tempfile
+import zipfile
+import requests
+
+# Commonmark GitHub flavored project
+COMMONMARK_GFM_URL = "https://github.com/github/cmark-gfm/archive/refs/heads/master.zip"
+SPEC_EXTRACTOR = "cmark-gfm-master/test/spec_tests.py"
+SPEC_FILES = ["spec.txt"]
+
+
+@functools.lru_cache(1)
+def commonmark_gfm_tests():
+    tests = []
+    r = requests.get(COMMONMARK_GFM_URL)
+
+    with tempfile.TemporaryDirectory() as tmpdir:
+        with tempfile.NamedTemporaryFile(mode="wb", suffix=".zip") as tfile:
+            tfile.write(r.content)
+
+            with zipfile.ZipFile(tfile.name, mode="r") as fzip:
+                fzip.extractall(tmpdir)
+
+        tmp_path = Path(tmpdir)
+        test_folder = (tmp_path / SPEC_EXTRACTOR).parent
+
+        sys.path.insert(0, str(test_folder))
+
+        from spec_tests import get_tests
+        from normalize import normalize_html
+
+        for testfile in SPEC_FILES:
+            tests.extend(get_tests(str(test_folder / testfile)))
+
+        sys.path.remove(str(test_folder))
+
+    return tests
+
+# Javascript function to render a markdown string `md`
+# to HTML using the JupyterLab parser.
+get_jupyterlab_rendered_markdown = """async (md) => {
+    const app = window.jupyterlab ?? window.jupyterapp;
+
+    const pluginId = '@jupyterlab/rendermime-extension:plugin';
+
+    const plugin = app._pluginMap[pluginId];
+
+    if (!plugin.activated) {
+        await app.activatePlugin(pluginId);
+    }
+
+    const renderer = plugin.service.createRenderer('text/markdown');
+    await renderer.renderModel(
+        plugin.service.createModel({ data: { 'text/markdown': md } })
+    );
+
+    return renderer.node.innerHTML;
+}"""