🤖 config comments: stop describing a vendoring pipeline this repo doesn't have
Same bug as cfc7683, in a place that audit's grep didn't reach — scoped to docs.njk, it missed eleventy.config.js entirely. Two comments ported verbatim from the website repo described mechanisms that are true there and false here: - The docs collection's comment said the corpus is "vendored from the platform repo by hack/sync-docs.mjs." Neither exists on this side, and none is planned — the corpus is authored by hand in src/docs/. Left uncorrected, the comment reads like a pointer to a script the next task's implementer would go looking for, or write from scratch. - The docsTitle filter's comment justified reading titles from raw markdown by saying "the vendored markdown carries no front matter — it is the platform repo's file, byte for byte." Carloku's markdown isn't vendored from anywhere; it carries no front matter because that's how its own corpus is authored, on purpose, so it reads the same in an editor, on GitHub, or fetched raw via docsmd.njk. Both rewritten to describe what's actually true on this side. The slugify/Go-Anchor comment and the fenceDelim/sameFence port comment are left untouched: they're two of the four things Task 1's brief named load-bearing and required kept with their comments, and unlike the two above they describe a real algorithm-parity requirement and a real code provenance fact, not corpus authorship or a verification mechanism that doesn't exist here.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
1 file changed,
+10
−6
eleventy.config.js+10 −6
diff --git a/eleventy.config.js b/eleventy.config.js| index 7cd7dca..05d185a 100644 |
| --- a/eleventy.config.js |
| +++ b/eleventy.config.js |
| @@ -60,9 +60,11 @@ export default function (eleventyConfig) { |
| ); |
| eleventyConfig.setLibrary("md", md); |
| - // The docs corpus is vendored from the platform repo by |
| - // hack/sync-docs.mjs. Pages are addressed by slug — "cli" — which is |
| - // also how nav.json and every internal /docs link name them. |
| + // The docs corpus is authored here, by hand, under src/docs/ — there |
| + // is no vendoring and no sync step. Pages are addressed by slug — |
| + // "cli" — which is also how src/_data/docsnav.json (the corpus's |
| + // hand-maintained table of contents) and every internal /docs link |
| + // name them. |
| eleventyConfig.addCollection("docs", (api) => |
| api |
| .getFilteredByGlob("src/docs/**/*.md") |
| @@ -124,9 +126,11 @@ export default function (eleventyConfig) { |
| eleventyConfig.addFilter("docsSlugify", slugify); |
| - // A page's own "# " title. The vendored markdown carries no front |
| - // matter — it is the platform repo's file, byte for byte — so the |
| - // title is read from the source rather than declared twice. |
| + // A page's own "# " title. The docs markdown carries no Eleventy front |
| + // matter — it's plain prose with a leading "# " heading, authored that |
| + // way on purpose so the corpus reads the same whether you're in an |
| + // editor, on GitHub, or fetched raw via docsmd.njk — so the title is |
| + // read from the source rather than declared twice. |
| eleventyConfig.addFilter("docsTitle", (raw) => { |
| for (const line of String(raw).split("\n")) { |
| const m = /^#\s+(.+?)\s*#*$/.exec(line); |