v3.15.1

Fixed

• Bot-protection challenge pages are reported, not extracted (#67, #68). A Cloudflare managed challenge (cf-mitigated: challenge) or a Bunny Shield proof-of-work page (cdn-challenge: true, or the same interstitial delivered with HTTP 200) used to surface either as an opaque Page fetch failed with status 403 or as the interstitial itself rendered to Markdown. /api now answers 502 with {"code":"bot_challenge","vendor":"..."} and the workaround (save the page in your browser, upload via POST /api/html) in the error text. Nothing lands in the cache or history. When only the headless render hits the challenge while the plain fetch got the real page, the static extraction is kept and extractorReason records why. PullMD deliberately does not attempt to solve these challenges. (#69, #70)

Documentation

• README opens with what PullMD does, not with its history: tagline, "Why PullMD" with a measured token table, highlights, quick start. The "What's new in v3" section is gone from the README; the changelog is the single record. (#66)

Full changelog: https://github.com/AeternaLabsHQ/pullmd/blob/main/CHANGELOG.md