Implement the full landing page for atlas.atomstorm.ai per the PRD at
Docs/Execution/Landing-Page-PRD-2026-03-14.md. Includes design spec,
bilingual Astro site, CI pipeline, and all assets.
Key deliverables:
- DESIGN.md: 7-section Atom Web Design specification
- Astro 5.x static site with 15 components and 11 page sections
- Bilingual i18n (zh/en) with path-based routing
- Build-time GitHub release manifest integration
- Release channel state machine (Stable/Prerelease/Coming Soon)
- CSS design tokens mapped from AtlasBrand.swift
- Self-hosted fonts (Space Grotesk, Instrument Sans, IBM Plex Mono)
- OG social sharing images for both locales
- GitHub Actions workflow for GitHub Pages deployment
- Zero client JS, 227KB page weight, 17/17 quality checks passed
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Add validation pass that checks all selected recovery items before any
restore operations begin. This prevents partial in-memory restore success
when a later item fails.
Map helper-backed restore destination conflicts to restore-specific
rejection paths instead of falling back to generic execution-unavailable
messages.
Bump version to 1.0.1 and update CHANGELOG with release notes.
ActionItem now carries optional targetPaths so plan.execute can use
plan-carried targets instead of reconstructing execution intent from
findings. This improves execution reliability and enables proper restore
mappings for recovery items.
- Add targetPaths field to ActionItem domain model
- Update plan execution to prefer plan-carried targets with finding fallback
- Expand safe cache path fragments for direct-trash execution
- Add gate review documentation for ATL-211/212/215
- Bump protocol version to 0.3.0
- update internal beta hardening docs and Smart Clean execution coverage for the 2026-03-16 hardening slice
- add pnpm store detection, execution fixtures, and regression coverage across adapters, infrastructure, and app model tests
- distinguish file-backed vs Atlas-only recovery messaging in History and restore summaries
- preserve completed Smart Clean recovery entries when later targets fail, and report failed findings explicitly instead of losing side effects
