What is the PostEverywhere MCP server and what can an AI agent do with it?

The PostEverywhere MCP server is the address an AI agent connects to so it can work inside your PostEverywhere account. Once connected, the agent can connect accounts, write and schedule posts, upload media, run campaigns, read analytics and manage webhooks, all through plain conversation.

How to connect#

1. In PostEverywhere, open Agents in the sidebar and pick your AI app. Each card shows the exact steps for that app.
2. Give the agent the server address: https://mcp.posteverywhere.ai. https://mcp.posteverywhere.ai/mcp works too.
3. Sign in when the agent opens the PostEverywhere window, check the account shown next to Signed in as, and click Approve.

If your AI app cannot open a sign-in window, it can use an API key instead. Create one on the Developers page and send it as the header Authorization: Bearer pe_live_.... See Getting an API key.

The connection uses the workspace that was selected when you approved it. See How AI agents and API keys work with workspaces.

Which address does my app use?#

App	Address	Tools
Claude Code, Cursor, Codex CLI, Gemini CLI, Windsurf, Zed, n8n, your own code	https://mcp.posteverywhere.ai	44
Claude (added from the Claude connector directory)	https://mcp.posteverywhere.ai/claude	43
ChatGPT (added from the ChatGPT app directory)	https://mcp.posteverywhere.ai/chatgpt	34

The Claude directory listing leaves out generate_image. The ChatGPT app leaves out ten tools: connect_credential_account, generate_image, generate_caption, get_brand, create_webhook, update_webhook, get_webhook, list_webhooks, delete_webhook and test_webhook. ChatGPT writes its own captions, the ChatGPT app does not take tools that collect passwords or make images, and the Brand Kit and webhook tools are newer and not in that app listing yet. If you add PostEverywhere to Claude or ChatGPT by hand with the main address, you get the full set.

What an agent can do#

Accounts list_accounts, get_account and get_account_health show what is connected and whether anything needs reconnecting. create_connect_link and create_reconnect_link give you a link to connect or repair an account in your browser. connect_credential_account connects Telegram, Discord, Bluesky or WordPress without a browser. get_me shows who is signed in, the workspace, the plan and what is left on each quota. get_platform_rules lists each platform's limits.

Posts create_post publishes now, schedules, or saves a draft for you to check. schedule_post publishes or schedules a draft. update_post, delete_post, get_post, list_posts and list_posts_advanced manage what is already there. bulk_create_posts makes many posts in one call. get_queue uses your posting queue and get_best_times picks a good time. get_post_results shows where a post ended up, and retry_failed_post or retry_failed_posts try again. approve_post and request_post_changes run the team approval flow; see Can an AI agent create drafts and send posts for approval?.

Media upload_media_from_url, create_media_upload and complete_media_upload bring images, videos and PDFs into your Media library. list_media, get_media and delete_media manage the library. generate_image and generate_caption use your AI credits to make an image or caption. See How does an AI agent upload images and videos?.

Campaigns list_campaigns, create_campaign, get_campaign, update_campaign and delete_campaign group posts into campaigns.

Analytics get_analytics_summary reports how your accounts and posts performed. get_best_times and get_post_results add detail.

Webhooks list_webhooks, create_webhook, get_webhook, update_webhook, delete_webhook and test_webhook set up HTTP calls when posts publish or fail. See Webhooks.

Brand get_brand reads your Brand Kit (what the business does, voice, rules, banned words, hashtags, colours) so the agent writes in your voice.

Good to know#

• Before a post is accepted, PostEverywhere checks for problems it can already see, such as an account that needs reconnecting, a caption that is too long for X (Twitter), or a daily platform limit. The agent gets the reason and a suggested fix, so nothing is scheduled that would fail.
• A post that a platform has published cannot be recalled by the agent or by PostEverywhere.
• Every call an agent makes is recorded on the Logs page. See Where can I see what my API keys and AI agents did?.
• Calls made by an agent count against the same API limits as any API key. See API rate limits.

Developer detail: MCP integration guide, Claude, other agents, authentication.