The Aether MCP server exposes every API endpoint as a tool that Claude can call directly. Instead of writing code, you describe what you want in natural language and Claude handles the API calls — including multi-step workflows like "check my inbox, find any unread DMs from verified accounts, and draft replies."
Installing the MCP server
The MCP server runs via npx — no global install required. Add it to your MCP client config:
Claude Desktop:
// Claude Desktop: ~/Library/Application Support/Claude/claude_desktop_config.json
// Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"aether": {
"command": "npx",
"args": ["-y", "aether-mcp"],
"env": {
"AETHER_API_KEY": "aet_live_your_key_here",
"AETHER_API_URL": "https://api.aetherhq.dev"
}
}
}
}Cursor:
# Cursor: ~/.cursor/mcp.json
{
"mcpServers": {
"aether": {
"command": "npx",
"args": ["-y", "aether-mcp"],
"env": {
"AETHER_API_KEY": "aet_live_your_key_here"
}
}
}
}Restart Claude Desktop or Cursor after updating the config. You should see an Aether tools indicator in the interface — hover it to see all available tools.
What you can do in natural language
Once connected, you can ask Claude things like:
- "Show me my social inbox — any unread messages from the last 24 hours?"
- "Schedule a post to Instagram and LinkedIn for tomorrow at 9am EST: [post text]"
- "What were my top 3 posts by engagement last week?"
- "When is the best time to post to my LinkedIn audience?"
- "Generate a connect link for Instagram so I can add my client's account"
- "Reply to the last comment on my Instagram account that asked about pricing"
Agent-optimized error messages
The MCP server is designed for LLM self-correction. When a tool call fails — for example, because a profile ID doesn't exist or a scheduled time is in the past — the error response includes a structured hint that Claude can act on without human intervention:
Building AI agent workflows
The MCP integration isn't just for interactive use. You can use Claude's computer use or extended thinking capabilities to build fully autonomous social media agents that:
- Monitor your inbox and auto-reply to common questions
- Generate and schedule content based on your company blog RSS feed
- Analyze your analytics and surface the highest-performing content patterns
- Cross-post content with platform-appropriate adaptations
The same MCP tools work in Cursor and Windsurf, so you can trigger social media actions directly from your code editor — useful for shipping announcements, for example.
Next steps
- Read the MCP server documentation for the full tool list
- Check the integrations page for n8n, Make, and Zapier workflows
- Get an API key from the dashboard — free tier supports 3 connected accounts