Daily Planner

A Python MCP server that generates a markdown morning briefing, orchestrated by a Copilot CLI agent skill.

The briefing includes: Outlook calendar events (via Work IQ), today's and tomorrow's Things tasks grouped by Area, AI action suggestions for unassigned tasks, and LLM-summarised repository activity from GitHub and Azure DevOps.

Requirements

• macOS (Things 3 database + Keychain)
• Python ≥ 3.12
• UV package manager
• GitHub Copilot CLI installed and authenticated (gh auth login + gh extension install github/gh-copilot)

Installation

git clone <your-repo-url> daily-planner
cd daily-planner
uv sync

MCP Server Setup

Copilot CLI needs to know how to start the daily-planner MCP server. Add it to your Copilot CLI MCP configuration.

Copilot CLI (~/.copilot/mcp-config.json)

Create or edit ~/.copilot/mcp-config.json:

{
  "mcpServers": {
    "daily-planner": {
      "type": "stdio",
      "command": "uv",
      "args": ["run", "--directory", "/absolute/path/to/daily-planner", "python", "-m", "daily_planner"]
    }
  }
}

Important: Replace /absolute/path/to/daily-planner with the actual absolute path to your clone of this repo (e.g. /Users/you/Source/daily-planner).

VS Code (.github/copilot/mcp.json)

This repo also includes a VS Code MCP config at .github/copilot/mcp.json for use within the editor. If you run the agent from within VS Code's Copilot Chat, it will use this config automatically — just make sure you open the project folder in VS Code.

Work IQ (optional — calendar events)

If you have access to the Microsoft Work IQ MCP server for Outlook calendar events, add it to the same mcp-config.json:

{
  "mcpServers": {
    "daily-planner": {
      "type": "stdio",
      "command": "uv",
      "args": ["run", "--directory", "/absolute/path/to/daily-planner", "python", "-m", "daily_planner"]
    },
    "workiq": {
      "type": "stdio",
      "command": "...",
      "args": ["..."]
    }
  }
}

Without Work IQ, the briefing will still generate — the calendar section will show "Calendar data unavailable" and all other sections will work normally.

Configuration

config/repos.txt

One repository per line. Lines starting with # are comments.

github:owner/repo
ado:org/project/repo

For example, to track the Radius project:

github:radius-project/radius

config/settings.toml

Customise the output path:

[output]
path = "~/Desktop"
repos_file = "config/repos.txt"

Authentication

GitHub — the token is resolved in this order:

1. GITHUB_TOKEN environment variable
2. gh auth token (if GitHub CLI is installed and authenticated via gh auth login)
3. macOS Keychain (service: daily-planner, account: github_access_token)

Azure DevOps — the token is resolved in this order:

1. ADO_TOKEN environment variable
2. az account get-access-token (if Azure CLI is installed and authenticated via az login)
3. macOS Keychain (service: daily-planner, account: ado_access_token)

The easiest path: just run gh auth login once and you're set for GitHub repos.

Usage

Copilot CLI agent (primary)

copilot --agent morning-briefing --allow-all-tools "Generate my morning briefing"

--allow-all-tools pre-approves MCP tool calls so the agent can run without interactive permission prompts.

The agent will:

1. Fetch calendar events from Work IQ (if configured)
2. Fetch today's and tomorrow's tasks from Things 3
3. Fetch repo activity from GitHub/ADO
4. Summarise repo activity using its LLM
5. Generate a two-page PDF on your Desktop

Output: ~/Desktop/2026-03-13 Friday.pdf (or your configured path)

Direct MCP server (for testing)

uv run python -m daily_planner

Troubleshooting

Problem	Solution
No such agent: morning-briefing	Ensure you're running from the repo root so Copilot CLI can find .github/agents/
Agent starts but immediately exits	Check that ~/.copilot/mcp-config.json exists and the --directory path is correct
"Things database not found"	Ensure Things 3 is installed; check ~/Library/Group Containers/
"Token expired" for GitHub	Run gh auth login to refresh your CLI token, or update GITHUB_TOKEN env var
"Repos config file not found"	Create config/repos.txt per the configuration section above
PDF sections show "Unavailable"	Check stderr output for the specific integration error
Font sizes not changing	Ensure config/settings.toml is valid TOML syntax

Debug Logging

Set DAILY_PLANNER_DEBUG=1 to capture a detailed JSONL log of every tool invocation, API call, and error during a briefing run:

DAILY_PLANNER_DEBUG=1 uv run python -m daily_planner

The log file is written to the configured output directory as debug_YYYY-MM-DD_HHMMSS_<pid>.jsonl. Each line is a JSON object with fields:

• timestamp — ISO 8601
• level — DEBUG, INFO, ERROR
• operation — e.g. get_today_tasks, github.fetch_commits
• message — human-readable description
• direction — request, response, or internal
• data — structured payload (truncated at 5,000 chars)
• duration_ms — elapsed time for response entries
• traceback — full traceback on errors

Debug logging writes only to a file (never stdout) so it cannot interfere with MCP stdio transport.

Development

# Run tests
uv run pytest

# Lint
uv run ruff check src/ tests/

# Security audit
uv run pip-audit

License

See LICENSE.