anymail

by @davidjonas in AI & LLM

# Install this skill:

npx skills add davidjonas/AnyMail --skill "anymail"

Install specific skill from multi-skill repository

# Description

Operate the AnyMail email CLI in E:\work\AnyMail to manage Gmail IMAP/SMTP profiles, auth (App Passwords via keyring), inbox listing/searching/reading, flags (seen/star/archive/trash), replying, sending email (with attachments), and auditing via invocation logs. Use when you need deterministic, scriptable email operations from the host (Windows) via the `anymail` CLI, including JSON/pipe workflows and troubleshooting with `anymail doctor`.

# SKILL.md

name: anymail
description: Operate the AnyMail email CLI in E:\work\AnyMail to manage Gmail IMAP/SMTP profiles, auth (App Passwords via keyring), inbox listing/searching/reading, flags (seen/star/archive/trash), replying, sending email (with attachments), and auditing via invocation logs. Use when you need deterministic, scriptable email operations from the host (Windows) via the anymail CLI, including JSON/pipe workflows and troubleshooting with anymail doctor.

Ground rules

Treat AnyMail as the single source of truth for email operations.
Prefer read-only commands first (inbox/search/read) before send/flag.
Prefer machine output:
Use --json when available.
Use --pipe only when you need UIDs for chaining.

Where it lives

Project root: E:\work\AnyMail
Skill scripts: skills/anymail/scripts/

Recommended execution path (most reliable)

Use the wrapper script to run AnyMail with the project venv when present:

python skills/anymail/scripts/anymail_run.py -- <anymail args...>

Examples:
- python skills/anymail/scripts/anymail_run.py -- doctor
- python skills/anymail/scripts/anymail_run.py -- profile list --json
- python skills/anymail/scripts/anymail_run.py -- inbox personal --unread --limit 20 --json

The wrapper:
- Uses .venv\\Scripts\\python.exe if it exists
- Otherwise falls back to python on PATH
- Runs -m anymail.cli from the repo (so you don’t rely on an installed console script)

Common workflows

1) Setup / verify account (profile)

Add profile:
... -- profile add <name> --email [email protected]
Store app password (interactive):
... -- auth set <name>
Verify connectivity:
... -- auth status --profile <name> --json

2) Triage inbox (safe, read-only)

List unread:
... -- inbox <profile> --unread --limit 50 --json
Search:
... -- search <profile> --subject "invoice" --since 90d --json
Read:
... -- read <profile> <uid> --headers --json
... -- read <profile> <uid> --body --json

3) Save attachments

List:
... -- read <profile> <uid> --attachments list --json
Save:
... -- read <profile> <uid> --attachments save --out <dir> --json

4) Mutate state (ask/confirm in chat before doing)

Mark read/unread:
... -- flag <profile> <uid> --seen true|false --json
Star/unstar:
... -- flag <profile> <uid> --star true|false --json
Archive/trash:
... -- flag <profile> <uid> --archive --json
... -- flag <profile> <uid> --trash --json

5) Send email (be deliberate)

Dry run first:
... -- send <profile> --to [email protected] --subject "Hi" --body "..." --dry-run
Then send:
... -- send <profile> --to ... --subject ... --body ... [--cc ...] [--bcc ...] [--attach <path>] --json

Monitoring / debugging

Health check:
... -- doctor
Invocation logs (audit what happened):
... -- logs list --since 24h --json
... -- logs query --command inbox --limit 100 --json

If output looks wrong

Re-run with --json.
Run doctor.
Inspect logs for errors (logs list --outcome error --since 7d --json).

Reference

Read skills/anymail/references/cli.md for the command map + JSON shapes.

# Supported AI Coding Agents

This skill is compatible with the SKILL.md standard and works with all major AI coding agents:

⚡ Amp 🚀 Antigravity 🤖 Claude Code 🦀 Clawdbot 📝 Codex ▶️ Cursor 🤖 Droid 💎 Gemini CLI 🐙 GitHub Copilot 🪿 Goose 📊 Kilo Code 🔧 Kiro CLI 💻 OpenCode 🦘 Roo Code 🌲 Trae 🏄 Windsurf

Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.