guten-morgen 0.23.0

CLI for Morgen calendar and task management — optimised for LLM consumption

guten-morgen

A CLI for Morgen calendar and task management, designed for both humans and LLM agents.

All commands emit structured JSON, making it easy to pipe into scripts, jq, or feed directly to AI coding assistants like Claude Code.

Features

• Unified task view — see tasks from Morgen, Linear, and Notion in one place
• Calendar groups — filter events by work/personal/family with a single flag
• Time-blocking — schedule tasks as calendar events with tasks schedule
• Tag lifecycle — model task stages (Active, Waiting-On, Someday) with tags
• LLM-friendly output — --json, --response-format concise, --jq, --fields for token-efficient responses
• Smart caching — TTL-based cache with cache clear and cache stats
• Bearer token auth — auto-detects Morgen desktop app for 5x higher rate limits

Installation

Requires Python 3.10+.

# Run without installing (uv)
uvx guten-morgen --help

# Install globally (uv — recommended)
uv tool install guten-morgen

# Install globally (pipx)
pipx install guten-morgen

# Install into current environment (pip)
pip install guten-morgen

# From source (development)
git clone https://github.com/tenfourty/guten-morgen.git
cd guten-morgen
uv sync --all-extras
uv run pre-commit install

All methods expose both gm (short) and guten-morgen (long) commands.

Setup

1. Get an API key from Morgen Platform (Settings > API Keys)

2. Create config (choose one):

   # Interactive setup (recommended)
   gm init
   
   # Or manually: project-local config
   cp config.toml.example guten-morgen.toml
   # Edit guten-morgen.toml and add your api_key
   

   Config discovery: $GM_CONFIG → guten-morgen.toml (walks up from CWD) → ~/.config/guten-morgen/config.toml

3. (Optional) Install Morgen desktop app for 5x higher API rate limits. gm auto-detects the desktop app's auth token and uses it when available. No configuration needed.

4. Verify it works:

   gm accounts
   gm today --json
   

Quick Start

# What's coming up?
gm next --json --response-format concise

# Full daily overview
gm today --json

# List overdue tasks across all sources
gm tasks list --status open --overdue --json --group-by-source

# Create and time-block a task
gm tasks create --title "Write design doc" --due 2026-02-20 --duration 90
gm tasks schedule <task-id> --start 2026-02-20T10:00:00

# Filter by tag
gm tasks list --tag "Active" --status open --json

Run gm --help for the full command reference.

Calendar Groups

Groups let you filter events by context. Configure in guten-morgen.toml:

default_group = "work"
active_only = true

[groups.work]
accounts = ["[e-mail]:google"]
calendars = ["Work Calendar"]

[groups.personal]
accounts = ["[e-mail]:fastmail"]
calendars = ["Personal"]

Use --group personal to switch context, or --group all to see everything.

Global Options

Option	Description
--format table|json|jsonl|csv	Output format (default: table)
--json	Shortcut for --format json
--fields <list>	Select specific fields
--jq <expr>	jq filtering on output
--response-format concise	~1/3 the tokens (great for LLMs)
--short-ids	Truncate IDs to 12 chars
--group NAME	Filter by calendar group
--no-cache	Bypass cache

Development

# Install dev dependencies
uv sync --all-extras
uv run pre-commit install

# Run tests
uv run pytest -x -q --cov

# Type checking
uv run mypy src/

# Lint
uv run ruff check .

Pre-commit hooks enforce ruff, mypy, bandit, pytest (90% coverage minimum), and ggshield secret scanning. To use ggshield, create a free account and set GITGUARDIAN_API_KEY.

Architecture

src/guten_morgen/
  cli.py        Click commands — boundary layer (model -> dict)
  client.py     MorgenClient — typed API wrapper (Pydantic models)
  models.py     Pydantic v2 models
  output.py     Render pipeline (table/json/jsonl/csv + fields + jq)
  errors.py     Exception hierarchy -> structured JSON on stderr
  config.py     XDG config discovery + API settings
  auth.py       Bearer token auth via Morgen desktop app
  time_utils.py Date range helpers
  cache.py      TTL-based request cache
  groups.py     Calendar group filtering from guten-morgen.toml
  retry.py      Rate-limit retry with dual-mode countdown

The boundary rule: client.py returns Pydantic models, cli.py converts with model_dump(), output.py only sees dicts.

Claude Code Integration

This project includes a CLAUDE.md with conventions and a .claude/ directory with hooks and skills for use with Claude Code. These are optional — the CLI works without them.

License

Apache 2.0