For AI AgentsA Review Gate for
Your AI AgentsYour agent drafts. A human approves. Automate It publishes. Connect anything that speaks MCP and let it work your content queue—without ever handing it your social credentials.
How It WorksAgents Propose. People Approve.The same review gate that keeps slop out of your feeds keeps your agents honest.
01
Connect Your AgentPoint any MCP client at the remote server—it signs you in with OAuth right in the browser. Headless agents use the ait CLI with a scoped API key instead.
02
It Drafts ContentThe agent claims or creates a task, adds content items, and marks the task complete—posts, threads, articles, or podcast episodes.
03
A Human ApprovesSupervised tasks wait in the review queue for your decision—approve, reject, or send back with feedback. Give your agent a viewer-scoped key and review is always enforced.
04
Automate It PublishesThe platform holds the social credentials and posts the approved content to X, LinkedIn, Instagram, and every other connected destination.
ConnectTwo Ways to Plug InCoding assistants connect over MCP. Autonomous agents get a CLI and an AgentSkills-format skill. Same tools, same review gate.
Coding Assistants · MCP
Claude Code
Codex
Grok Build
Cursor
Claude Code, Codex, Grok Build, and Cursor connect over the remote MCP server and sign in with OAuth right in the browser—no token to manage. Once connected, your assistant gets all 38 tools for tasks, content, files, and automations.
terminal
claude mcp add automate-it \
--transport http \
https://api.automate.it.com/mcp# your client opens the browser to sign in via OAuth
Autonomous Agents · CLI + Skill
OpenClaw
Hermes
OpenClaw and Hermes agents install the ait CLI from npm or the Agent Skill from ClawHub, then authenticate headlessly with a scoped API key and work the queue through the same review gate. Mint the key in the web app on your Profile screen, under Developer API Keys → Create Key.
terminal
npm install -g automate-it
set -a; . ~/.automate-it.env; set +a # AUTOMATE_IT_API_KEY=ak_...ait task create --claim \
--title"Launch thread"--output-types x
GuardrailsBuilt So You Can SleepAutonomy for your agents, control for you.
Scoped API KeysGrant only what the agent needs—tasks:write and content:write for a submitter, tasks:read for a watcher. Revoke and reissue anytime.
Credentials Stay HomeYour agent never sees your social tokens. It submits content; the platform is the only thing that ever talks to X, Meta, or LinkedIn.
Roles on Top of ScopesWorkspace roles still apply. A viewer-role key can propose work but never approve it or skip review—only keys you grant a reviewer or admin role can waive the gate.
Full Audit TrailEvery agent action—claims, submissions, approvals—lands in the workspace activity feed with the acting identity attached.
Give Your Agents a Quality BarStart your 7-day free trial, create a scoped API key, and put your first agent behind the review gate.