Translations update from Hosted Weblate

conf.py

@@ -83,6 +83,18 @@
     "sphinx_search.extension",
     "rstjinja",
     "myst_parser",
+    "llms_txt",
+]
+
+# llms.txt: an index of every page for coding agents (docs/llms_txt.py)
+llms_txt_description = (
+    "Adafruit CircuitPython - Python for microcontrollers. "
+    "Each page is also served as Markdown with the Accept: text/markdown header."
+)
+llms_txt_exclude = [
+    "docs/porting",
+    "docs/pdf",
+    "shared-bindings/support_matrix",
 ]
 
 # Add any paths that contain templates here, relative to this directory.

docs/llms_txt.py

@@ -0,0 +1,82 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 Vladimir Smitka
+#
+# SPDX-License-Identifier: MIT
+
+"""Write llms.txt, an index of the docs for LLM agents (https://llmstxt.org/).
+
+It reuses what the HTML build already parsed: one line per page with its title
+and first paragraph, in table-of-contents order. Read the Docs serves each page
+as Markdown on request, so the index links the normal HTML pages.
+"""
+
+from docutils import nodes
+from sphinx import addnodes
+
+DESCRIPTION_LENGTH = 160
+
+
+def page_order(env, docname, seen):
+    """Yield docnames depth-first in table-of-contents order."""
+    if docname in seen:
+        return
+    seen.add(docname)
+    yield docname
+    for child in env.toctree_includes.get(docname, []):
+        yield from page_order(env, child, seen)
+
+
+def is_skipped(node):
+    """Raw HTML, parameter lists and notes; an API description counts."""
+    if isinstance(node, (nodes.raw, nodes.field_list, nodes.comment)):
+        return True
+    # Sphinx derives API descriptions from Admonition too, so keep those.
+    return isinstance(node, nodes.Admonition) and not isinstance(node, addnodes.desc)
+
+
+def first_paragraph(doctree):
+    """Plain text of the first paragraph worth using as a description."""
+    for paragraph in doctree.findall(nodes.paragraph):
+        parent = paragraph.parent
+        while parent is not None and not is_skipped(parent):
+            parent = parent.parent
+        if parent is not None:
+            continue
+        text = " ".join(paragraph.astext().split())
+        if text:
+            if len(text) > DESCRIPTION_LENGTH:
+                text = text[: DESCRIPTION_LENGTH - 3].rstrip() + "..."
+            return text
+    return ""
+
+
+def write_llms_txt(app, exception):
+    if exception is not None or app.builder.name not in ("html", "dirhtml"):
+        return
+    env = app.env
+    excluded = set(app.config.llms_txt_exclude)
+    lines = ["# " + app.config.project, ""]
+    if app.config.llms_txt_description:
+        lines += ["> " + app.config.llms_txt_description, ""]
+    lines += ["## Pages", ""]
+    seen = set()
+    ordered = list(page_order(env, app.config.root_doc, seen))
+    ordered += sorted(env.found_docs - seen)
+    for docname in ordered:
+        if docname in excluded or docname not in env.titles:
+            continue
+        title = env.titles[docname].astext().strip()
+        description = first_paragraph(env.get_doctree(docname))
+        link = app.builder.get_target_uri(docname)
+        entry = "- [{}]({})".format(title, link)
+        if description:
+            entry += ": " + description
+        lines.append(entry)
+    with open(app.outdir / "llms.txt", "w", encoding="utf-8") as f:
+        f.write("\n".join(lines) + "\n")
+
+
+def setup(app):
+    app.add_config_value("llms_txt_description", "", "html")
+    app.add_config_value("llms_txt_exclude", [], "html")
+    app.connect("build-finished", write_llms_txt)
+    return {"version": "1.0", "parallel_read_safe": True, "parallel_write_safe": True}