AI Agents

Social media API built for AI agents

Aether exposes every social media capability as a native MCP tool — post, schedule, analytics, inbox, webhooks. Claude, Cursor, and any MCP-compatible agent can act on 7 platforms without integration code.

What AI agents can do with Aether

Real use cases that work today — no hypothetical AI futures.

Scheduling

Autonomous content scheduling

Agent generates content, selects the optimal posting time based on analytics, and schedules across platforms — without human intervention. Claude calls aether_post_create with scheduledFor set to the best-performing window.

Inbox

AI-powered inbox management

Agent monitors incoming DMs and comments via webhooks, classifies intent, drafts responses with Claude, and calls aether_inbox_reply — closing support loops without a human in the loop.

Analytics

Automated analytics reporting

Agent pulls weekly account metrics across all platforms, synthesizes trends, flags anomalies, and posts a summary — to Slack, email, or as a scheduled post itself. No dashboard needed.

Campaigns

Campaign orchestration

Agent receives a campaign brief, generates platform-specific copy with overrides per platform, uploads media, and schedules the full cross-platform launch sequence — one prompt, seven platforms.

Onboarding

Connect link delivery

Agent generates OAuth connect links for new client onboarding, sends them via email or Slack, monitors for connection completion, and confirms account is ready — end-to-end client setup without manual steps.

Engagement

Real-time engagement response

Agent subscribes to comment.created webhooks. When a comment matches a trigger keyword, it generates a response, checks brand guidelines, and replies — within seconds of the comment appearing.

MCP tools exposed by Aether

Every tool below is available to any MCP-compatible agent. Claude Desktop, Cursor, Windsurf, and custom agent frameworks using the MCP SDK can call these directly — no REST wiring required.

ToolWhat it does
aether_post_createCreate or schedule a post to one or more platforms
aether_post_listList posts by status, platform, or date range
aether_post_cancelCancel a scheduled post before it publishes
aether_analytics_post_metricsGet engagement metrics for a specific post
aether_analytics_account_metricsGet follower and engagement trends for an account
aether_inbox_listFetch unread DMs and comments across all platforms
aether_inbox_replyReply to a DM or comment
aether_connect_link_createGenerate an OAuth connect link for a platform
aether_webhook_createRegister a webhook endpoint for real-time events

Quick start

Add Aether to Claude Desktop in 2 minutes

One config block. Restart Claude Desktop. Every Aether tool is immediately available in the tool palette — no code, no build step.

~/.claude/claude_desktop_config.json
# 1. Add to ~/.claude/claude_desktop_config.json
{
  "mcpServers": {
    "aether": {
      "command": "npx",
      "args": ["-y", "aether-mcp"],
      "env": {
        "AETHER_API_KEY": "sk_live_..."
      }
    }
  }
}

# 2. Restart Claude Desktop. Aether tools appear in the tool palette.

# 3. Prompt Claude naturally:
# "Post our product launch announcement to Instagram, LinkedIn, and Threads at 9am tomorrow"
# "Show me our top-performing Instagram posts from last month"
# "Check the inbox and draft replies to any unanswered DMs"
# "Generate a connect link for the new client's Instagram account"

Works the same way for Cursor and Windsurf — see the full MCP server setup guide.

Build your own AI agent with the REST API

Prefer full control? Combine Aether's REST API with the Anthropic SDK directly. Webhook triggers, Claude reasoning, Aether actions — all in one loop.

inbox-agent.ts
import Aether from "aether";
import Anthropic from "@anthropic-ai/sdk";

const aether = new Aether({ apiKey: process.env.AETHER_API_KEY });
const claude = new Anthropic();

// AI agent: monitor inbox, draft AI response, auto-reply
async function handleInboxWithAI() {
  const messages = await aether.inbox.list({ status: "unread", limit: 10 });

  for (const msg of messages.data) {
    const response = await claude.messages.create({
      model: "claude-opus-4-8",
      max_tokens: 200,
      messages: [{
        role: "user",
        content: `You are a helpful brand assistant. Reply to this social media message in under 50 words.
Platform: ${msg.platform}
From: ${msg.from.name}
Message: ${msg.text}`,
      }],
    });

    const replyText = response.content[0].type === "text"
      ? response.content[0].text
      : "";

    await aether.inbox.reply(msg.id, { text: replyText });
    await aether.inbox.markRead(msg.id);
  }
}

Free tier · 3 accounts · no credit card

Get your free API key

Frequently asked questions

What is an AI social media agent?+

An AI social media agent is an autonomous system that can post content, read analytics, manage inboxes, and respond to events across social platforms without manual intervention. It combines an AI model (like Claude) with a social media API (like Aether) and optionally an MCP server that exposes API endpoints as callable tools the AI can invoke directly.

How does Aether's MCP server work?+

Aether ships a Node.js MCP server (aether-mcp) that exposes every REST endpoint as a callable tool following the Model Context Protocol. When you add it to Claude Desktop or Cursor's MCP config, the AI can directly call Aether tools — post to Instagram, pull analytics, reply to DMs — as part of its reasoning process, without you writing any integration code.

Can I build an autonomous social media agent with Claude and Aether?+

Yes. The typical pattern: subscribe to webhooks for real-time triggers (new comment, new DM, scheduled time), pass the event payload to Claude for analysis and response generation, then call the Aether REST API or MCP tools to act on Claude's output. The aether-mcp server makes this even simpler — Claude can call tools natively without you writing the REST integration layer.

What social platforms does the AI agent support?+

Instagram, TikTok, LinkedIn, Facebook, YouTube, Threads, and Reddit — all 7 platforms Aether covers. The AI agent can post, schedule, read analytics, and manage inbox messages across all platforms from a single set of MCP tools or REST endpoints.

Is the MCP server free to use?+

Yes. The aether-mcp package is open-source and free. You need an Aether API key (free tier available) and a compatible AI agent environment like Claude Desktop, Cursor, or Windsurf. No additional charges for MCP usage beyond your normal Aether plan.

Your agent. Every platform. One API.

MCP server · REST API · webhooks · 7 platforms · free tier.

Get your free API key →