diff --git a/.github/workflows/pages.yml b/.github/workflows/pages.yml
new file mode 100644
index 0000000..3bab95e
--- /dev/null
+++ b/.github/workflows/pages.yml
@@ -0,0 +1,28 @@
+name: Pages
+on: push
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/configure-pages@v3
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 20
+          cache: npm
+          cache-dependency-path: prototypes/basic/package-lock.json
+      - name: Setup hq
+        run: |
+          curl https://github.com/ludovicianul/hq/releases/download/hq-1.3.2/hq_linux_amd64_1.3.2.tar.gz --output - | tar -xv
+      - name: Build site
+        working-directory: ./prototypes/basic
+        run: |
+          npm install
+          coproc npm run dev &
+          while ! (echo >/dev/tcp/localhost/3000); do sleep 1; done
+          export PATH:../../:$PATH
+          make -C ../../doc -j all
+          make -C ../../doc -j all
+      - name: Upload artifact
+        uses: actions/upload-pages-artifact@v2
diff --git a/doc/Makefile b/doc/Makefile
new file mode 100644
index 0000000..795bc88
--- /dev/null
+++ b/doc/Makefile
@@ -0,0 +1,49 @@
+HQ ?= $(shell which hq)
+
+SOURCES := $(wildcard ../lib/importer/assets/docs/*.html)
+OBJECTS := $(subst ../lib/importer/assets/docs/,./,${SOURCES})
+
+# Create a list of all of the assets the pages need, by looking at the scripts
+# and stylesheets they intended to download.
+assets.txt: ${SOURCES}
+	(echo $^ | xargs -n1 -IINPUT ${HQ} -r '[data-plugin-only]' -f INPUT -a src 'script[src]'; \
+	 echo $^ | xargs -n1 -IINPUT ${HQ} -r '[data-plugin-only]' -f INPUT -a href 'link[rel=stylesheet]') | \
+		sort -u | grep . > $@
+
+# Now build a Makefile dependency include, specifying the correct download
+# location for each asset and that we require it for a complete package.
+assets.mk: assets.txt
+	for ASSET in $$(cat assets.txt); do \
+		export SRC=$$(echo $$ASSET | sed 's:^[^\/]:/plugin-assets/%40register-dynamics%2Fimporter/assets/docs/.:'); \
+		printf "all: $$(basename $$ASSET)\n"; \
+		printf "%s: SRC=%s\n\n" $$(basename $$ASSET) $$SRC; \
+	done > $@
+
+.PHONY: all
+all: ${OBJECTS} | assets.txt # Rewrite all of the asset paths to be local now.
+	for ASSET in $$(cat assets.txt); do \
+		sed -i '' -e "s:$$ASSET:$$(basename $$ASSET):" ${OBJECTS}; \
+	done
+include assets.mk
+
+# We can generate the .gitignore from all of the things we know we are going to
+# end up with in the root dir.
+.gitignore: assets.txt
+	(cat $^; printf '%s\n' ${OBJECTS}; echo assets.mk) | xargs basename > $@
+all: .gitignore
+
+# Download the page and remove any elements marked only for plugin use.
+%.html: ../lib/importer/assets/docs/%.html
+	curl --fail -sL http://localhost:3000/$(subst ../lib/importer/,/plugin-assets/%40register-dynamics%2Fimporter/,$^) | \
+		${HQ} -r '[data-plugin-only]' -o $@ html
+
+# Download the asset. This is the easiest way to get the real thing, as some of
+# them (e.g. application.css) are generated by the Prototype Kit and can't just
+# be copied from the disk.
+%.css %.js:
+	curl --fail -sL "http://localhost:3000/${SRC}" > $@
+
+.PHONY: clean
+clean:
+	-cat .gitignore | xargs $(RM)
+	-$(RM) assets.mk assets.txt