Buffer MCP server for your AI

No. Buffer connects with a personal API key, so there is no OAuth app, client secret, or redirect URL to set up. Generate one in the Buffer app under Settings → API → Personal Keys → + New Key, which is available on every plan, then paste it into the Personal API key field. Whatever permissions you give the key are the permissions the agent inherits, so a read-only key stays read-only no matter what you ask for.

Every channel already connected to the Buffer account that owns the key: Twitter/X, Instagram, Facebook, LinkedIn, TikTok, YouTube, Pinterest, Threads, Bluesky, Mastodon, Google Business and the others Buffer supports. list_channels enumerates them for an organization, and get_channel returns a single channel's per-day posting schedule along with the actions it allows. Connecting a brand new channel is still done in Buffer itself, though a Channel connected trigger fires when it happens.

No. create_post targets one channel, so a five-channel announcement is five calls, each with its own copy and timing, which is why an agent usually runs list_channels first. daily_posting_limits reports how many posts each channel has already sent or scheduled against its daily quota for a given day, so you can see where there is room before queueing. There is also a threshold trigger for a channel using up its daily allowance.

get_post_metrics returns one post's likes, comments, shares, reach, impressions, clicks and engagementRate, and get_post returns a post with its assets, tags, metrics and any publishing error attached. For reporting rather than spot checks, aggregated_post_metrics rolls metrics up across an organization over a date range and can be filtered to specific channels or tags. That aggregate call is the one to point an agent at when you ask for a weekly or monthly recap.

16 operations across 6 modules, covering Posts, Ideas, Channels, Account and Metrics, split 11 read to 5 write. 13 load into your client by default and the rest stay searchable, so the tool list your AI sees does not balloon. There are also 10 triggers PopMCP can watch, including post published, post failed to publish, draft created, channel disconnected and a channel hitting its daily limit.

The graphql tool runs any Buffer GraphQL query or mutation directly, including schema introspection and custom field selections, so nothing in Buffer's schema is out of reach. One naming quirk catches people out: Buffer's mutation for changing an existing post is editPost, not updatePost, which is why the curated tool is edit_post. Both edit_post and delete_post work on a post id, and that id comes from list_posts.

No. PopMCP hosts it and gives you a Streamable HTTP MCP URL. There is nothing to deploy, no callback URL to register, and no infrastructure to keep running or patch.

Any client that supports a remote Streamable HTTP MCP server — Claude Desktop, Claude Web, ChatGPT custom connectors, Codex and the Codex CLI, OpenCode and Z.ai's ZCode, as well as editors such as Cursor, Windsurf and Zed. Your client signs in to PopMCP with OAuth; the provider credential never leaves the server.

Loading 16 tool definitions into every conversation burns context before the work starts. PopMCP loads a curated 13-tool set covering the everyday workflows and keeps the rest searchable, so your AI pulls a specialist tool only when a task needs it. You can switch a connection to the full catalog at any time.

A tool is something your AI calls — there are 16 here. A trigger is something PopMCP watches for you: 10 Buffer events that can start work on the PopMCP side. Triggers never enter your AI client's context, and none of them ask you to paste a callback URL into Buffer.