Gmail Checker

A prioritized Gmail digest that filters out promotions and noise to surface urgent and personal emails.

Install
cmdop skills install agensi-gmail-checker

Gmail Checker is a skill that connects to the Google Gmail API to fetch unread messages from the last 24 hours, or a configurable custom duration. It applies a multi-tier filtering system driven by a configurable JSON definition to sort incoming mail into HIGH, MEDIUM, and LOW priority categories based on domains and keywords the user defines. Social media notifications, promotions, automated updates, forum digests, and marketing messages are filtered out before any output is produced, reducing the volume of low-value content passed to an LLM.

The skill produces two output formats: a human-readable formatted summary showing sender, subject line, and priority level, and a structured JSON object suitable for agent-to-agent workflows. This dual output makes it usable both in interactive developer environments and in headless server pipelines.

The skill is described as compatible with Claude Code, Cursor, and OpenClaw, and supports both headless server environments and local developer machines. It is implemented with Python 3.x.

This is the right choice when an agent needs to act on genuinely important email — such as security alerts or personal correspondence — without consuming context budget on automated inbox noise. It is not a general-purpose email client and does not support sending, replying to, or archiving messages based on the available facts.

Use cases

  • Fetch and triage unread Gmail messages inside an AI agent workflow
  • Filter out promotional and automated emails before passing inbox content to an LLM
  • Generate a daily digest of high-priority personal and business email
  • Feed structured email priority data into agent-to-agent pipelines via JSON output
  • Run inbox monitoring on a headless server without a graphical environment
  • Define custom domain and keyword rules to tune priority classification

When to use it

  • An agent needs to surface only urgent or personal emails and ignore inbox noise
  • A pipeline requires structured JSON email data for downstream agent processing
  • Running in a headless server environment where a full email client is unavailable
  • Context window budget is limited and low-value email must be excluded before LLM processing

When not to use it

  • The workflow requires sending, replying to, or archiving emails — this skill only reads
  • Access to email providers other than Gmail is needed
  • Real-time push-based email monitoring is required rather than periodic polling
  • No Google Gmail API credentials or access are available