Index documents as well

[hrstoyanov]

What problem does this solve?

I have codebase along with docs in htmlx, asciidoc, mdx - can we index the docs as well an correlate with the code graph?

Proposed solution

No idea

Alternatives considered

No response

Confirmations

• I searched existing issues and this is not a duplicate.

[DeusData]

Great idea, thank you @hrstoyanov 🙏 This is adjacent to a couple of open threads: #504 (import existing ADR docs) and #346 (a memory.md). The graph today is code structure first, so correlating prose docs with code is a larger feature.

The realistic first step is lighter than full doc indexing: extract headings and the code symbols a doc references (fenced code identifiers, inline code spans, explicit links) and create "documents" edges to existing graph nodes, rather than semantically embedding the prose. Which formats matter most to you, in priority order (Markdown and MDX, AsciiDoc, HTML)? That'll help us scope a first cut!

[hrstoyanov]

I have a web scraper tool that downloads documentation (html) and converts it to markdown.

You may want to take a look at QMD - it does very well with the docs, not so much with figuring out code and graphs

Ideally I want one tool that does it all

[hrstoyanov]

Btw I love the "single fast executable" approach. Too much models and python slop out there