The SKILL.md body loads into context on every activation, so one-time install/ setup prose was wasted context once emcli is running. Move it out: - New AGENTIC-MANUAL.md: get-the-files bootstrap, binary install (incl. options and build-from-source, folding in the old references/install.md), EMCLI_KEY, account discovery. Fetched only during first-time setup. - SKILL.md trimmed (182→~145 lines) to the recurring path: security model, a short "Files & first run" pointer + per-session preflight, the list/get/ack/send workflow, JSON envelope, command table, enforcement, do/don't. - Remove references/install.md (folded in); fix RELEASING.md pointer. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
emcli
A single command-line program that mediates all email access for an AI agent. The agent never
holds your email password and never talks to the mail server directly — every read and send goes
through emcli, which enforces the limits you configure (read-only/read-write, sender and
recipient whitelists, subject filters). Even with faulty instructions, the agent can't read mail
it isn't permitted to see or send mail to people it isn't permitted to contact.
Getting started
export EMCLI_KEY="$(head -c 32 /dev/urandom | base64)" # one-time: generate & save a key
emcli init # create the DB, add your first account
emcli doctor # confirm it connects and authenticates
Documentation
See the User Manual for full setup, account configuration (including Gmail app passwords), the agent and admin command reference, the JSON output format, and troubleshooting.
For AI agents, skills/emcli is an
Agent Skill that teaches an agent to read and send mail through emcli,
including a binary installer.