v0.1.0

Changelog

v0.1.0 - 2026-04-25

Added

• AI weekly summary generation with provider support for OpenAI-compatible APIs and Gemini.
• Audio generation for weekly summaries using Gemini TTS or ElevenLabs.
• Optional PDF extracted-text audio generation with minimum length threshold.
• Two-phase processing pipeline: first generate all summaries, then generate audio.
• Job artifact tracking in files/ai_jobs/ with input/output markdown files and manifest.jsonl.
• Resume mode improvements for missing audio generation based on existing outputs and manifest state.
• Gemini TTS health reporting (GREEN/YELLOW/RED) and status indicators in the Web UI.

Changed

• PDF text extraction is now embedded directly into week pages instead of requiring separate pages.
• External downloadable link fallback now uses inline source warnings instead of creating separate note pages.
• Gemini TTS throttling defaults and adaptive backoff tuning to reduce rate-limit failures.
• Web UI now exposes full AI/TTS controls, PDF audio presets, and richer run statistics.

Fixed

• Resolved PCM audio decoding issues by wrapping Gemini L16 PCM responses into WAV.
• Fixed duplicate filename collisions for mirrored files and generated audio assets.
• Corrected parse_float scope issue in Web UI backend.
• Improved retry handling for 429/5xx errors, including Retry-After support.