v2.2.0 - embeddings generation job

What's Changed

• Embeddings generation job (bge-m3, en+es chunks) by @jaycollett in #4

Full Changelog: v2.1.0...v2.2.0

v2.1.0 - chapter-marker condensing

Adds a chapter-marker condensing job alongside the sermon AI-content job.

• New POST /submit_chapters + GET /chapters_status/<guid> and a dedicated chapters worker thread (own table, idempotent UPSERT, output normalization).
• generate_chapters() condenses English transcription timings into 5-12 bilingual (EN / Mexican-Spanish) chapter markers via the Claude CLI.

Consumed by the SermonOrchestrator GenerateChapters step, which forwards the markers to the web app's /api/sermon_chapters.

v2.0.2: Lower CLI effort to medium

Drops --effort from high to medium to stretch the $200/month Agent SDK credit. Equivalent quality for single-shot structured-output sermon summaries.

v2.0.1: Claude Code CLI replaces OpenAI

Replaces OpenAI chat completions with the Claude Code CLI subprocess (claude -p, model claude-opus-4-7, --max-turns 2). Bundles Node.js + Claude Code CLI into the image. Fixes worker.py to set updated_at on error path so the 24h cleanup actually applies.

0.1.4

Adds logic to purge old jobs from the database.

0.1.3

Updates number of key quotes

0.1.2

Updates the number of key quotes.

0.1.1

Adds workflow

0.1.0

Initial commit