v0.4.8 — AI Advisor error handling and Docker release images

What's Changed

AI Advisor

• Added GEMINI_MODEL support so self-hosters can override the default Gemini model when provider or account compatibility requires it.
• Backend AI/provider failures now surface as user-safe 502 AI_PROVIDER_ERROR responses while preserving technical diagnostics in backend logs.
• Frontend API error handling now preserves structured backend error payloads so chat and auth flows show accurate messages.
• The chat UI no longer misclassifies provider or model failures as "AI Advisor Disabled".

Self-Hosting and Docker

• Published multi-arch Docker Hub release images from semver tags:
  • slsadmin/bucketwise-planner-backend:0.4.8
  • slsadmin/bucketwise-planner-frontend:0.4.8
• Fixed frontend Docker build ordering so workspace dependencies resolve correctly during image builds.
• Managed deployments such as CasaOS can keep using DB_SCHEMA_MODE=manual to avoid routine restarts applying schema changes automatically.
• Self-hosting and AI docs now call out the required backend recreation step after AI env var changes.
• Republished the 0.4.8 Docker images to include curl in the backend and frontend runtime images so Compose health checks no longer fail unhealthy on missing binaries. Related issue: #33.

Upgrading

No database migration is required for v0.4.8.

If you change backend AI env vars such as AI_ENABLED, GEMINI_API_KEY, or GEMINI_MODEL, recreate the backend container with:

docker compose up -d --force-recreate backend

docker compose restart backend is not sufficient for those env changes.

• Self-hosting guide: https://github.com/solid-logic-studios/bucketwise-planner/blob/main/docs/SELF_HOSTING.md
• Upgrade guide: https://github.com/solid-logic-studios/bucketwise-planner/blob/main/docs/UPGRADING.md

---

Issue reference: #33

Full Changelog: v0.4.6...v0.4.8