chk
/
journal-bot
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
21 Commits 1 Branch 0 Tags 146 KiB
Python 72.4%
C# 24.7%
PowerShell 2.9%
Go to file
beo3000 d2f8319772 fix(process): derive target date from message send day, not processing day 
A message sent yesterday but processed today (e.g. by the next service
run) landed in today's daily note. The send time was captured in
received_at but only HH:MM reached the LLM; the date defaulted to
date.today(). Now today/weekday in the processor context are derived per
item from received_at, so "gestern"/"heute" resolve relative to when the
message was actually written. Adds day_context() helper and a regression
test (sent 2026-06-25, processed 2026-06-26 -> note 2026-06-25).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
 		2026-06-26 08:19:29 +02:00
docs/superpowers feat: windows service + WPF tray GUI for journal-bot 2026-06-16 09:37:39 +02:00
scripts feat: windows service + WPF tray GUI for journal-bot 2026-06-16 09:37:39 +02:00
src/journal_bot fix(process): derive target date from message send day, not processing day 2026-06-26 08:19:29 +02:00
tests fix(process): derive target date from message send day, not processing day 2026-06-26 08:19:29 +02:00
tools feat(tray): use antialiased multi-size B icon (Brain) 2026-06-16 10:24:02 +02:00
.env.example chore: bootstrap journal-bot project skeleton 2026-06-15 15:01:04 +02:00
.gitignore chore: bootstrap journal-bot project skeleton 2026-06-15 15:01:04 +02:00
CLAUDE.md feat: windows service + WPF tray GUI for journal-bot 2026-06-16 09:37:39 +02:00
README.md chore: bootstrap journal-bot project skeleton 2026-06-15 15:01:04 +02:00
pyproject.toml test(prompt): regression cases for date resolution, meta filter, clarification 2026-06-15 17:42:39 +02:00
uv.lock chore: bootstrap journal-bot project skeleton 2026-06-15 15:01:04 +02:00

README.md

Journal Bot

Telegram → Whisper → LM Studio / Claude → Obsidian Daily Note.

Setup

  1. uv sync --extra dev
  2. cp .env.example .env und Werte eintragen
  3. LM Studio starten, Modell qwen/qwen3-vl-8b laden

Commands

  • python -m journal_bot ingest — Telegram abholen, Queue füllen
  • python -m journal_bot process — Queue mit LM Studio verarbeiten
  • python -m journal_bot both — Ingest + Process in einem Lauf
  • python -m journal_bot write --target-path PATH --entry-file FILE — Helper für Claude-Skill

Siehe Spec: ../brain/docs/superpowers/specs/2026-06-14-telegram-journal-bot-design.md