Cordia app previews and Hub workflow status.
-
Added Cordia Hub profile onboarding with region selection, public Hub profile publishing, verified handle lookup, profile retry handling, and Hub-backed profile links after setup.
-
Added generated app live previews with an embedded project app panel, fullscreen viewing, trusted Cordia app URL handling, and UI controls that keep released apps visible beside project discussion.
-
Added a Forum project type for admin-created community projects that do not require generated GitHub repositories, Cloudflare Pages setup, release targets, or improvement releases.
-
Improved Cordia release and app provisioning with shared project prefixes, generated repo team access, Cloudflare Pages DNS/domain setup, awaited bootstrap sync, generated template readiness checks, and isolated app setup error prompts.
-
Expanded Cordia discovery and SEO with page metadata, robots and llms files, a richer web manifest, public project activity cues, footer branding, corrected Cordia copy, and Google tag review coverage.
-
Fixed project discussion and detail UI behavior for nested reply releases, creator avatar stability, last release failure display, reply z-index, improvement spacing, project panel hover states, and admin user/project counts.
-
Added agent invocation workflow status publishing from the harness through Hub daemon updates so long-running implementation, remediation, and conflict-resolution work can report live run metadata.
-
Improved the local dashboard by replacing Chart.js charts with D3 donut charts, adding source icons inside chart legends, hiding unavailable dictation controls, and refining logo and PR monitor UI polish.
-
Added live app URL tracking in task logs and snapshots so generated app preview links can flow through the broker into task detail views.
-
Kept retrying agent work out of parent failure state, removed duplicate Hub task generation paths, and added a distinct completion sound for auto-merged pull requests.
-
Improved skill validation diagnostics by returning the skill name, payload format, allowed, required, and optional parameters in invalid skill activation errors and A2A agent cards.