This release ships a new helleze config command for post-init reconfiguration, an improved init flow with a cleaner provider/model picker, and an unlocked managed model that respects the client's model choice. The landing page and changelog now share a single layout with mobile fixes, corrected copy, and SEO improvements. GitHub release notes no longer include a redundant version/date heading.
New
- New
helleze configcommand for post-init reconfigurationA newhelleze configcommand lets you interactively change your AI provider, model, and site URL after the initial setup, syncing changes back to the server. Theinitflow also now always asks for provider and model upfront, and private-repo users get a cleaner "Generate with" picker instead of a plain warning. 010150b
Improved
- Managed model proxy now respects your model choiceThe helleze-managed LLM proxy no longer pins a single model — it now accepts any model from the supported allowlist (defaulting to Sonnet), so your model preference set during init or config is actually used. 010150b
- Shared marketing layout, mobile fixes, copy corrections, and SEO improvementsThe landing page and changelog now share a single layout component so their header, nav, and footer can't drift apart. Mobile layouts are fixed down to 250px wide, overclaims in the copy have been corrected, and the site now includes JSON-LD structured data, Open Graph/Twitter cards, a sitemap, and RSS autodiscovery. 470243c
- GitHub release notes no longer include a redundant version/date headingThe
# v1.2.0 — 2026-06-10heading that appeared at the top of generated GitHub release note bodies has been removed, since GitHub already displays the tag and date in its own UI. 7da8701