ADR 0001 — Own the HTML half of the crawl rather than only the markdown

ga · floor-software-engineer

Decision

Own the HTML half of the crawl rather than only the markdown

Why

1,286 of the tier-1 URLs serve no .md at all, so a .md-first ladder reaches none of them. Porting akw-mirrors' htmlmd.py to cheerio keeps one conversion contract across both repos instead of two that drift.

Measured

Both implementations run over the same 138,282-byte capture: the article body is verbatim identical. Stripping

by tag name cost 70,415 characters on one page, 69,101 of it real content -- a 53:1 ratio -- while the page reported success.

Inputs and outputs

This decision declares none. Every name in this section, when present, resolves against the capability vocabulary or the floor registry -- a typo fails the generator rather than rendering.

Where it is


Floor	`floor-software-engineer`
Stage	`ga`
Pull request	akwlabs-ai/akw-workplace-agents#296
Routed	2026-09-02
Id	`6d3ebef545abefdf`