afterpython 0.2.0__tar.gz → 0.3.0__tar.gz

{afterpython-0.2.0 → afterpython-0.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: afterpython
-Version: 0.2.0
+Version: 0.3.0
 Summary: CLI tool to streamline Python package maintenance and generate a dedicated project website
 Keywords: package-maintenance,project-website,static-site-generator,developer-tools,documentation,cli-tool,ci/cd,mystmd,semantic-release,python-packaging
 Author: Stephen Yau
@@ -17,16 +17,17 @@ Classifier: Programming Language :: Python :: 3.13
 Requires-Dist: click>=8.2.1
 Requires-Dist: trogon>=0.6.0
 Requires-Dist: httpx>=0.28.1
-Requires-Dist: tomlkit>=0.13.3
-Requires-Dist: ruamel-yaml>=0.18.16
-Requires-Dist: gitpython>=3.1.45
-Requires-Dist: python-dotenv>=1.2.1
-Requires-Dist: pyproject-metadata>=0.10.0
-Requires-Dist: mystmd>=1.8.0
-Requires-Dist: pdoc>=16.0.0
-Requires-Dist: ruff>=0.14.14
-Requires-Dist: pre-commit>=4.5.1
-Requires-Dist: commitizen>=4.12.1
+Requires-Dist: tomlkit>=0.14.0
+Requires-Dist: ruamel-yaml>=0.19.1
+Requires-Dist: gitpython>=3.1.49
+Requires-Dist: python-dotenv>=1.2.2
+Requires-Dist: pyproject-metadata>=0.11.0
+Requires-Dist: jupytext>=1.19.1
+Requires-Dist: mystmd>=1.9.0
+Requires-Dist: great-docs>=0.9
+Requires-Dist: ruff>=0.15.12
+Requires-Dist: pre-commit>=4.6.0
+Requires-Dist: commitizen>=4.13.10
 Requires-Python: >=3.11
 Project-URL: homepage, https://afterpython.afterpython.org
 Project-URL: repository, https://github.com/AfterPythonOrg/afterpython
@@ -36,11 +37,13 @@ Description-Content-Type: text/markdown
 # AfterPython: Python Package Maintenance Toolkit and Project Website Generator
 
 [![afterpython](https://afterpython.org/shield.svg)](https://afterpython.org)
-![PyPI downloads](https://img.shields.io/pypi/dm/afterpython)
-[![PyPI](https://img.shields.io/pypi/v/afterpython.svg)](https://pypi.org/project/afterpython)
-![PyPI - Support Python Versions](https://img.shields.io/pypi/pyversions/afterpython)
+[![Downloads](https://static.pepy.tech/badge/afterpython/month)](https://pepy.tech/project/afterpython)
+[![PyPI](https://img.shields.io/pypi/v/afterpython.svg?cacheSeconds=300)](https://pypi.org/project/afterpython)
+![PyPI - Support Python Versions](https://img.shields.io/pypi/pyversions/afterpython?cacheSeconds=300)
 [![Discussions](https://img.shields.io/badge/Discussions-Let's%20Chat-green)](https://github.com/AfterPythonOrg/afterpython/discussions)
 [![Ask DeepWiki](https://deepwiki.com/badge.svg)](https://deepwiki.com/AfterPythonOrg/afterpython)
+[![View Code Wiki](https://assets.codewiki.google/readme-badge/static.svg)](https://codewiki.google/github.com/afterpythonorg/afterpython?utm_source=badge&utm_medium=github&utm_campaign=github.com/afterpythonorg/afterpython)
+<!-- ![PyPI downloads](https://img.shields.io/pypi/dm/afterpython) -->
 <!-- [![Pixi Badge](https://img.shields.io/endpoint?url=https://raw.githubusercontent.com/prefix-dev/pixi/main/assets/badge/v0.json)](https://pixi.sh) -->
 
 
@@ -50,7 +53,7 @@ Description-Content-Type: text/markdown
 [pre-commit]: https://pre-commit.com
 [pagefind]: https://pagefind.app
 [SemVer]: https://semver.org
-[pdoc]: https://pdoc.dev/
+[great-docs]: https://posit-dev.github.io/great-docs/
 [GitHub Actions]: https://github.com/features/actions
 [PyTorch]: https://pytorch.org
 [Scikit-Learn]: https://scikit-learn.org
@@ -63,7 +66,7 @@ Description-Content-Type: text/markdown
 
 ## Problem
 Going from **writing Python code to publishing and maintaining a package** is **time-consuming**.
-First-time maintainers must learn multiple tools and concepts, e.g. [MyST], [SemVer], CI/CD ([pre-commit] hooks, GitHub workflows), and they often spending hours just to deploy a documentation site whereas well-resourced projects like [PyTorch] and [Scikit-Learn] have **dedicated websites** that serve as their project hubs.
+First-time maintainers must learn multiple tools and concepts, e.g. [MyST], [SemVer], CI/CD ([pre-commit] hooks, GitHub workflows), and they often spend hours just to deploy a documentation site whereas well-resourced projects like [PyTorch] and [Scikit-Learn] have **dedicated websites** that serve as their project hubs.
 
 
 ## Solution
@@ -78,8 +81,8 @@ First-time maintainers must learn multiple tools and concepts, e.g. [MyST], [Sem
 - [x] Go from writing to **website deployment in minutes** — no need to learn any of the underlying tools
 - [x] Centralize all your content in a modern, **unified project website** — from documentation to blog posts
 - [x] Zero-config orchestration — Pre-configured modern tooling with sane defaults (see [Tech Stack](#tech-stack)), so you can start maintaining packages immediately **without learning each tool**
+- [x] **⚡ Full-text search** across **ALL** your content in your website — docs, blogs, tutorials, everything
 - [ ] Export content as PDF — for example, combine all blog posts into a single PDF file
-- [ ] **⚡ Full-text search** across **ALL** your content in your website — docs, blogs, tutorials, everything
 - [ ] **🤖 Embedded AI Chatbot** that answers questions directly using an in-browser LLM — at no cost
 
 ---
@@ -121,7 +124,7 @@ ap tui
 - [project-website-template]
 - [pre-commit]
 - [GitHub Actions]
-- [pdoc]
+- [great-docs]
 - [commitizen]
 - [uv]
 - [ruff]

{afterpython-0.2.0 → afterpython-0.3.0}/README.md

@@ -1,11 +1,13 @@
 # AfterPython: Python Package Maintenance Toolkit and Project Website Generator
 
 [![afterpython](https://afterpython.org/shield.svg)](https://afterpython.org)
-![PyPI downloads](https://img.shields.io/pypi/dm/afterpython)
-[![PyPI](https://img.shields.io/pypi/v/afterpython.svg)](https://pypi.org/project/afterpython)
-![PyPI - Support Python Versions](https://img.shields.io/pypi/pyversions/afterpython)
+[![Downloads](https://static.pepy.tech/badge/afterpython/month)](https://pepy.tech/project/afterpython)
+[![PyPI](https://img.shields.io/pypi/v/afterpython.svg?cacheSeconds=300)](https://pypi.org/project/afterpython)
+![PyPI - Support Python Versions](https://img.shields.io/pypi/pyversions/afterpython?cacheSeconds=300)
 [![Discussions](https://img.shields.io/badge/Discussions-Let's%20Chat-green)](https://github.com/AfterPythonOrg/afterpython/discussions)
 [![Ask DeepWiki](https://deepwiki.com/badge.svg)](https://deepwiki.com/AfterPythonOrg/afterpython)
+[![View Code Wiki](https://assets.codewiki.google/readme-badge/static.svg)](https://codewiki.google/github.com/afterpythonorg/afterpython?utm_source=badge&utm_medium=github&utm_campaign=github.com/afterpythonorg/afterpython)
+<!-- ![PyPI downloads](https://img.shields.io/pypi/dm/afterpython) -->
 <!-- [![Pixi Badge](https://img.shields.io/endpoint?url=https://raw.githubusercontent.com/prefix-dev/pixi/main/assets/badge/v0.json)](https://pixi.sh) -->
 
 
@@ -15,7 +17,7 @@
 [pre-commit]: https://pre-commit.com
 [pagefind]: https://pagefind.app
 [SemVer]: https://semver.org
-[pdoc]: https://pdoc.dev/
+[great-docs]: https://posit-dev.github.io/great-docs/
 [GitHub Actions]: https://github.com/features/actions
 [PyTorch]: https://pytorch.org
 [Scikit-Learn]: https://scikit-learn.org
@@ -28,7 +30,7 @@
 
 ## Problem
 Going from **writing Python code to publishing and maintaining a package** is **time-consuming**.
-First-time maintainers must learn multiple tools and concepts, e.g. [MyST], [SemVer], CI/CD ([pre-commit] hooks, GitHub workflows), and they often spending hours just to deploy a documentation site whereas well-resourced projects like [PyTorch] and [Scikit-Learn] have **dedicated websites** that serve as their project hubs.
+First-time maintainers must learn multiple tools and concepts, e.g. [MyST], [SemVer], CI/CD ([pre-commit] hooks, GitHub workflows), and they often spend hours just to deploy a documentation site whereas well-resourced projects like [PyTorch] and [Scikit-Learn] have **dedicated websites** that serve as their project hubs.
 
 
 ## Solution
@@ -43,8 +45,8 @@ First-time maintainers must learn multiple tools and concepts, e.g. [MyST], [Sem
 - [x] Go from writing to **website deployment in minutes** — no need to learn any of the underlying tools
 - [x] Centralize all your content in a modern, **unified project website** — from documentation to blog posts
 - [x] Zero-config orchestration — Pre-configured modern tooling with sane defaults (see [Tech Stack](#tech-stack)), so you can start maintaining packages immediately **without learning each tool**
+- [x] **⚡ Full-text search** across **ALL** your content in your website — docs, blogs, tutorials, everything
 - [ ] Export content as PDF — for example, combine all blog posts into a single PDF file
-- [ ] **⚡ Full-text search** across **ALL** your content in your website — docs, blogs, tutorials, everything
 - [ ] **🤖 Embedded AI Chatbot** that answers questions directly using an in-browser LLM — at no cost
 
 ---
@@ -86,7 +88,7 @@ ap tui
 - [project-website-template]
 - [pre-commit]
 - [GitHub Actions]
-- [pdoc]
+- [great-docs]
 - [commitizen]
 - [uv]
 - [ruff]

{afterpython-0.2.0 → afterpython-0.3.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "afterpython"
-version = "0.2.0"
+version = "0.3.0"
 description = "CLI tool to streamline Python package maintenance and generate a dedicated project website"
 license = "Apache-2.0"
 authors = [
@@ -35,19 +35,19 @@ dependencies = [
     "click>=8.2.1",
     "trogon>=0.6.0",
     "httpx>=0.28.1",
-    "tomlkit>=0.13.3",
-    "ruamel-yaml>=0.18.16",
-    "gitpython>=3.1.45",
+    "tomlkit>=0.14.0",
+    "ruamel-yaml>=0.19.1",
+    "gitpython>=3.1.49",
     # "pygithub>=2.8.1",
-    "python-dotenv>=1.2.1",
-    "pyproject-metadata>=0.10.0",
-    # "jupytext>=1.18.1",
+    "python-dotenv>=1.2.2",
+    "pyproject-metadata>=0.11.0",
+    "jupytext>=1.19.1",
     # "jinja2>=3.1.6",
-    "mystmd>=1.8.0",
-    "pdoc>=16.0.0",
-    "ruff>=0.14.14",
-    "pre-commit>=4.5.1",
-    "commitizen>=4.12.1",
+    "mystmd>=1.9.0",
+    "great-docs>=0.9",
+    "ruff>=0.15.12",
+    "pre-commit>=4.6.0",
+    "commitizen>=4.13.10",
     # "python-semantic-release>=10.5.2",
 ]
 
@@ -64,9 +64,15 @@ pcu = "afterpython.main:run_pcu"
 # override-dependencies = ["platformdirs>=4.5.0", "nodeenv>=1.9.1"]
 
 [build-system]
-requires = ["uv_build>=0.9.26,<0.10.0"]
+requires = ["uv_build>=0.11.8,<0.12.0"]
 build-backend = "uv_build"
 
 [tool.uv.build-backend]
 module-name = "afterpython"
 module-root = "src"
+
+[tool.uv.sources]
+afterpython = { path = ".", editable = true }
+
+[tool.pyright]
+typeCheckingMode = "basic"

{afterpython-0.2.0 → afterpython-0.3.0}/src/afterpython/builders/__init__.py

@@ -1,17 +1,25 @@
 from afterpython.builders.content_json import build_content_json
+from afterpython.builders.faq_json import build_faq_json
 from afterpython.builders.index_md import (
     create_placeholder_index_md_files,
     delete_placeholder_index_md_files,
 )
 from afterpython.builders.jupyter_notebook import build_jupyter_notebooks
+from afterpython.builders.llms_txt import build_llms_txt
+from afterpython.builders.marimo_notebook import build_marimo_readme
 from afterpython.builders.markdown import build_markdown
 from afterpython.builders.metadata import build_metadata
+from afterpython.builders.url_md import build_url_md
 
 __all__ = (
     "build_content_json",
+    "build_faq_json",
     "build_jupyter_notebooks",
+    "build_llms_txt",
+    "build_marimo_readme",
     "build_markdown",
     "build_metadata",
+    "build_url_md",
     "create_placeholder_index_md_files",
     "delete_placeholder_index_md_files",
 )

afterpython-0.3.0/src/afterpython/builders/faq_json.py

@@ -0,0 +1,43 @@
+from __future__ import annotations
+
+import json
+
+import click
+
+import afterpython as ap
+from afterpython._io.yaml import read_yaml
+
+build_path = ap.paths.build_path
+
+
+def build_faq_json():
+    """Build faq.json from the user's afterpython/faq.yml.
+
+    Output shape: a flat list of {question, answer, category}. `category` is
+    always present — empty string when unset — so the frontend can rely on a
+    uniform item shape. Order is preserved so the frontend can derive category
+    order from first appearance.
+    """
+    faq_yml_path = ap.paths.afterpython_path / "faq.yml"
+
+    if not faq_yml_path.exists():
+        click.echo("No faq.yml found, skip building faq.json")
+        return
+
+    click.echo("Building faq.json...")
+
+    raw = read_yaml(faq_yml_path)
+    if raw is None:
+        items: list[dict] = []
+    else:
+        items = [
+            {
+                "question": str(item["question"]),
+                "answer": str(item["answer"]),
+                "category": str(item["category"]) if item.get("category") else "",
+            }
+            for item in raw
+        ]
+
+    with open(build_path / "faq.json", "w") as f:
+        json.dump(items, f, indent=2)

{afterpython-0.2.0 → afterpython-0.3.0}/src/afterpython/builders/jupyter_notebook.py

@@ -5,24 +5,11 @@ from pathlib import Path
 import click
 
 import afterpython as ap
+from afterpython.builders.marimo_notebook import _create_molab_url, _get_molab_badge
 from afterpython.const import CONTENT_TYPES
 from afterpython.tools.pyproject import read_metadata
 
 
-def _get_molab_badge() -> str:
-    return "https://marimo.io/molab-shield.svg"
-
-
-def _create_molab_url(github_url: str, content_path: Path):
-    """Create a molab URL for a given content type and notebook path.
-    Args:
-        github_url: str, e.g. "https://github.com/AfterPythonOrg/afterpython"
-        content_path: str, e.g. "tutorial/test.ipynb"
-    """
-    github_url = github_url.replace("https://github.com/", "github/")
-    return f"https://molab.marimo.io/{github_url}/blob/main/afterpython/{content_path.as_posix()}"
-
-
 def _read_notebook(notebook_path: Path, content_path: Path) -> dict | None:
     """Read and validate a Jupyter notebook.
 

afterpython-0.3.0/src/afterpython/builders/llms_txt.py

@@ -0,0 +1,248 @@
+# VIBE-CODED
+"""Build llms.txt — a single catalog of the site for AI agents.
+
+Output: `afterpython/_website/static/llms.txt`. SvelteKit's static adapter
+serves it at `/llms.txt` so agents (Claude, GPT, etc.) can fetch a curated
+overview of the project + a link list of every published page.
+
+Conforms to the llms.txt convention (https://llmstxt.org/): a markdown file
+with a project H1, optional context paragraphs, and `## Section` headings
+each containing a link list.
+
+Flow
+----
+build_llms_txt()                                      # entry point
+  ├─ _read_readme() + _standardize_readme()           # → header content
+  └─ for each content_type in ORDERED_CONTENT_TYPES:
+       └─ _list_pages(content_type)                    # → [(title, url, abstract)]
+            (skips entries whose source .md no longer exists — those were
+             auto-generated placeholder index.md files deleted in postbuild)
+       └─ _format_section(content_type, blurb, entries)
+            ├─ heading: SECTION_HEADINGS[content_type]
+            ├─ blurb:   project.description from afterpython/{type}/myst.yml
+            └─ entries: `- [Title](/type/slug.md): abstract`
+"""
+
+from __future__ import annotations
+
+import json
+import re
+
+import click
+
+import afterpython as ap
+from afterpython._typing import tContentType
+from afterpython.const import CONTENT_TYPES
+
+# A line whose entire content is a markdown image (badge), with optional
+# link wrapper and optional HTML-comment wrapper.
+#   matches: ![alt](url)
+#            [![alt](url)](link)
+#            <!-- ![alt](url) -->
+#            <!-- [![alt](url)](link) -->
+# Anchored ^...$ so inline references in body prose are NOT matched.
+_BADGE_LINE = re.compile(
+    r"""
+    ^\s*                # leading whitespace
+    (?:<!--\s*)?        # optional HTML comment open
+    \[?                 # optional link open  '['
+    !\[[^\]]*\]         # ![alt]
+    \([^)]+\)           # (image_url)
+    (?:\]\([^)]+\))?    # optional  ](link_url)
+    (?:\s*-->)?         # optional HTML comment close
+    \s*$                # trailing whitespace
+    """,
+    re.VERBOSE,
+)
+
+# Order in which content-type sections appear in llms.txt. Sections with zero
+# entries are skipped. Reasoning: an agent answering "how do I X?" should hit
+# concepts first (doc), then how-to (tutorial/guide), then examples, then
+# background (blog) — strongest signal first.
+ORDERED_CONTENT_TYPES: list[tContentType] = [
+    "doc",
+    "tutorial",
+    "guide",
+    "example",
+    "blog",
+]
+
+# One-line hint that sits between the README and the section list, telling
+# an LLM reading llms.txt that the URLs below are fetchable markdown sources
+# (the Mintlify / Anthropic-docs convention). Costs ~20 tokens and removes
+# any ambiguity about what fetching `/doc/quickstart.md` returns.
+LINK_NOTE = (
+    "> Every link below points to the raw markdown source of the page."
+    " Fetch any URL directly to read its content as markdown."
+)
+
+# Human-readable display name for each content type's `## ` heading in llms.txt.
+SECTION_HEADINGS: dict[tContentType, str] = {
+    "doc": "Documentation",
+    "tutorial": "Tutorials",
+    "guide": "Guides",
+    "example": "Examples",
+    "blog": "Blog",
+}
+
+# Defensive sync check: if a new content type is added to const.CONTENT_TYPES,
+# the developer must remember to slot it into ORDERED_CONTENT_TYPES and
+# SECTION_HEADINGS. Import-time failure beats a silently-missing section.
+assert set(ORDERED_CONTENT_TYPES) == CONTENT_TYPES, (
+    "ORDERED_CONTENT_TYPES is out of sync with const.CONTENT_TYPES"
+)
+assert set(SECTION_HEADINGS.keys()) == CONTENT_TYPES, (
+    "SECTION_HEADINGS is out of sync with const.CONTENT_TYPES"
+)
+
+
+def _read_readme() -> str:
+    """Read the project's README.md from repo root, verbatim. Empty string if
+    the README is missing (e.g. brand new project)."""
+    readme_path = ap.paths.user_path / "README.md"
+    if not readme_path.exists():
+        return ""
+    return readme_path.read_text(encoding="utf-8")
+
+
+def _standardize_readme(text: str) -> str:
+    """Clean the README before embedding it in llms.txt.
+
+    Currently strips badges (shields, download counters, etc.) from the
+    README's *header zone* — everything before the first `## ` heading —
+    since they're chrome for human readers and noise for LLMs.
+
+    The header-zone restriction protects content images that legitimately
+    appear later in the README (e.g. the `<picture>` website screenshot
+    after the first H2 — also untouched because it's HTML, not a markdown
+    image).
+
+    Future ideas if more cleanup is wanted: strip GitHub-flavored anchor
+    links, install commands, "table of contents" sections.
+    """
+    lines = text.split("\n")
+
+    # Boundary of the header zone: index of the first `## ` line, or end of
+    # file if no H2 exists.
+    boundary = next(
+        (i for i, line in enumerate(lines) if line.startswith("## ")),
+        len(lines),
+    )
+
+    cleaned_header = [line for line in lines[:boundary] if not _BADGE_LINE.match(line)]
+    rejoined = "\n".join(cleaned_header + lines[boundary:])
+
+    # After removing 9 consecutive badge lines, you can end up with a stack
+    # of blanks where they used to be. Collapse 3+ newlines to 2 (one blank).
+    return re.sub(r"\n{3,}", "\n\n", rejoined)
+
+
+def _read_section_blurb(content_type: tContentType) -> str | None:
+    """Pull the section's one-line blurb from `afterpython/{type}/myst.yml`'s
+    `project.description`.
+
+    Returns None if myst.yml is missing or has no description set — in which
+    case the section in llms.txt simply omits the `> blurb` line.
+    """
+    from afterpython._io.yaml import read_yaml
+
+    myst_yml = ap.paths.afterpython_path / content_type / "myst.yml"
+    if not myst_yml.exists():
+        return None
+
+    data = read_yaml(myst_yml) or {}
+    description = data.get("project", {}).get("description")
+    return str(description).strip() if description else None
+
+
+def _list_pages(content_type: tContentType) -> list[tuple[str, str, str | None]]:
+    """Return `[(title, url, abstract), ...]` for each page MyST built.
+
+    Reads MyST's `_build/site/content/*.json` files (which already have
+    title, slug, and frontmatter resolved — no need to re-parse source).
+    Skips entries whose source .md no longer exists; those were auto-generated
+    placeholder index.md files that `delete_placeholder_index_md_files`
+    removed at the start of postbuild.
+    """
+    content_dir = ap.paths.afterpython_path / content_type
+    myst_content_dir = content_dir / "_build" / "site" / "content"
+    if not myst_content_dir.exists():
+        return []
+
+    pages: list[tuple[str, str, str | None]] = []
+    for json_file in sorted(myst_content_dir.glob("*.json")):
+        try:
+            data = json.loads(json_file.read_text(encoding="utf-8"))
+        except json.JSONDecodeError:
+            continue
+
+        # Filter placeholder index.md entries: their source has been deleted.
+        location = data.get("location", "")
+        source_md = content_dir / location.lstrip("/")
+        if not source_md.exists():
+            continue
+
+        slug = data.get("slug")
+        if not slug:
+            continue
+
+        frontmatter = data.get("frontmatter", {})
+        title = frontmatter.get("title") or slug
+        abstract = frontmatter.get("abstract")
+
+        url = f"/{content_type}/{slug}.md"
+        pages.append((title, url, abstract))
+    return pages
+
+
+def _format_link_entry(title: str, url: str, abstract: str | None) -> str:
+    """One line of the link list.
+
+    With abstract:    `- [Title](url): abstract`
+    Without abstract: `- [Title](url)`
+    """
+    if abstract:
+        return f"- [{title}]({url}): {abstract.strip()}"
+    return f"- [{title}]({url})"
+
+
+def _format_section(
+    content_type: tContentType,
+    blurb: str | None,
+    entries: list[tuple[str, str, str | None]],
+) -> str:
+    """Format one `## Section`: heading, optional blurb, link list."""
+    parts = [f"## {SECTION_HEADINGS[content_type]}", ""]
+    if blurb:
+        parts.extend([f"> {blurb}", ""])
+    for title, url, abstract in entries:
+        parts.append(_format_link_entry(title, url, abstract))
+    return "\n".join(parts)
+
+
+def build_llms_txt() -> None:
+    """Generate `_website/static/llms.txt` — one catalog file for the site.
+
+    Runs at the end of `postbuild()` after MyST output is finalized and
+    placeholder index.md files have been removed.
+    """
+    readme = _standardize_readme(_read_readme())
+
+    sections: list[str] = []
+    total_entries = 0
+    for content_type in ORDERED_CONTENT_TYPES:
+        entries = _list_pages(content_type)
+        if not entries:
+            continue
+        blurb = _read_section_blurb(content_type)
+        sections.append(_format_section(content_type, blurb, entries))
+        total_entries += len(entries)
+
+    body = readme.rstrip()
+    if sections:
+        body = body + "\n\n" + LINK_NOTE + "\n\n" + "\n\n".join(sections) + "\n"
+
+    output = ap.paths.website_path / "static" / "llms.txt"
+    output.parent.mkdir(parents=True, exist_ok=True)
+    output.write_text(body, encoding="utf-8")
+    click.echo(f"Generated {output} ({total_entries} entries)")