MCP App Store
by KeenEthics
Publer icon

Publer MCP Server for ChatGPT

By Kalemi

Overview

Turn ChatGPT into your social media manager with Publer. Create and schedule posts, manage your content calendar, analyze performance, research competitors, generate reports, and more by simply asking ChatGPT. Publer gives ChatGPT secure access to your actual social media context, from your content and calendar to performance and competitor data. So ChatGPT doesn’t just give you generic advice. It can understand what’s happening, recommend what to do next, and do it for you. All across Instagram, Facebook, LinkedIn, TikTok, X, Threads, Pinterest, YouTube, Google Business, Bluesky, Mastodon, Telegram, and WordPress. Posts are saved as drafts unless you ask to schedule or publish them, and nothing is deleted until you confirm.

How to connect

  1. ChatGPT: Open the Publer listing in the ChatGPT app directory and choose Connect. ChatGPT listing Publer exposes 54 tools here.

Tools

analyze_recent_performance

ChatGPT
Fetch recent posts and analytics, then return a compact audit summary with streaming progress.

change_publer_post_state

ChatGPT
Promote a draft to scheduled, or pull a scheduled post back to a draft, without re-creating it. The post keeps its existing scheduled time — set the time first with reschedule_publer_post if it needs a new one. state "scheduled" queues the post to go out, so it requires user_confirmed:true after the user says yes in a NEW message; moving back to a draft does not. To publish immediately instead, use publish_publer_draft. Only the member who created the post can change its state.

change_publer_workspace

ChatGPT
Switch the locked workspace. Omit workspace to re-show the list of names. Do not switch silently.

confirm_delete_publer_label

ChatGPT
Permanently delete a label after the user explicitly confirmed in a NEW chat message (the confirmation card's Delete button calls this itself). Requires a prior delete_publer_label preview for the same label: pass the same label and user_confirmed:true. Never call this in the same turn as delete_publer_label.

confirm_delete_publer_posts

ChatGPT
Permanently delete posts after the user explicitly confirmed in a NEW chat message (the confirmation card's Delete button calls this itself). Requires a prior delete_publer_posts preview for the same posts: pass the same refs you passed to the preview (none when the posts were attached) and user_confirmed:true. Never call this in the same turn as delete_publer_posts. Never ask the user for ids.

create_publer_ideas

ChatGPT
Save account-less content ideas in Publer Ideas; an idea is never posted to any network. Only when the user asks for an idea. A draft for an account or network is submit_publer_posts with when:"draft" — never save an idea because no account was named; look the account up or ask which one.

create_publer_photo_draft

ChatGPT
Use when the user attached or generated an image/video in chat and wants a Publer draft. Prefer create_publer_post_from_file (file param) on ChatGPT. REQUIRED: top-level file or media object — ChatGPT injects via openai/fileParams. Never pass /mnt/data paths, media_url with sandbox paths, media_path, or media_base64. Creates a Draft (not scheduled/published), dated ~1 hour ahead.

create_publer_post_from_file

ChatGPT
PRIMARY tool when the user attaches or generates an image/video in ChatGPT and wants a Publer draft/post. REQUIRED top-level file argument — ChatGPT's connector rewrites sandbox paths (/mnt/data/...) into { download_url, file_id } via openai/fileParams. Do NOT use submit_publer_posts with media_url or nested paths for chat files (that causes: File arg rewrite paths are required when proxied mounts are present). Creates a Draft by default. Set publish:true only when the user explicitly asks to publish.

delete_publer_label

ChatGPT
PREVIEW only — never deletes. Call this once when the user asks to delete a label. It draws a confirmation card with the warning and the buttons, so do NOT restate it: say one short line and stop. Never call confirm_delete_publer_label in the same turn — the card does that itself when the user confirms. To take a label off one post, use update_publer_post with remove_labels instead; deleting removes it everywhere.

delete_publer_posts

ChatGPT
PREVIEW only — never deletes. Call this once when the user asks to delete posts/drafts, and call it with NO refs when the posts they mean are the ones already attached. It draws a confirmation card that names every post and carries the buttons, so do NOT list the posts in your reply and do NOT ask the question again in text: say one short line and stop. Never call it twice in a turn, and never call confirm_delete_publer_posts in the same turn — the card does that itself when the user confirms.

export_publer_analytics

ChatGPT
Make a downloadable CSV or PDF report of analytics and return its download link. Use when the user asks to export, download or get a report/file of their analytics — for numbers to read in chat, use the other analytics tools. Covers every account in the workspace unless account is passed (or exactly one account is attached). Publer runs one export per user at a time. Share the download_url as a markdown link; if status is working, the file is still being made — check later with get_publer_job_status.

generate_campaign_plan

ChatGPT
Generate a multi-day campaign plan from a brief using recent Publer performance signals.

get_analytics_chart_data

ChatGPT
Fetch time-series data for one or more analytics charts. Pass chart ids from list_analytics_charts (e.g. followers, post_reach, post_engagement — not bare "reach"/"engagements"). Each chart includes summary.total for the period — use that for headlines; do not treat a trailing week with value 0 as the period total. Omit account for workspace aggregate where supported.

get_best_times_to_post

ChatGPT
Return the best days and times to post for your audience.

get_competitors_analytics

ChatGPT
Return paginated analytics for competitor accounts.

get_hashtag_insights

ChatGPT
Return hashtag performance analytics for a workspace or account.

get_hashtag_performing_posts

ChatGPT
Return top performing posts for a specific hashtag.

get_members_analytics

ChatGPT
Return per-member posting activity and engagement metrics.

get_post_insights

ChatGPT
Return each post's engagement (likes, comments, shares…) in a date range — for the user's own accounts, or for their tracked competitors with competitors:true. Sort by likes to get the best-performing posts first.

get_publer_job_status

ChatGPT
Poll async Publer job status after draft or media upload, or an analytics export that came back working. Pass the job value returned. A finished export carries download_url — share it as a markdown link.

get_publer_user

ChatGPT
Return the name and plan of the Publer user this connection signs in as. The email is included only when the user asks for it.

get_recent_posts

ChatGPT
Recent posts of the user's own accounts, with NO card. This is the lookup you use to analyse, summarise, compare or reason about their posts from the last days days up to today — posts dated in the future are not included; competitors' posts come from get_post_insights with competitors:true. When the user asked to SEE their posts or drafts, call list_publer_posts or list_publer_drafts instead so the card is drawn. Each post includes a ref number for update/delete. Never show ids to the user.

label_publer_media

ChatGPT
Add or remove labels on files in the media library. add_labels / remove_labels change those labels and keep the rest; labels replaces the file's whole set ([] clears it). Pass the files as media_ids — ref.id from list_publer_media / lookup_publer_media; with none, the files the user selected in the media card are used. A new name creates the label, so check list_publer_labels first and reuse the user's spelling.

list_analytics_charts

ChatGPT
List available analytics charts. Use each chart's id with get_analytics_chart_data chart_ids.

list_competitors

ChatGPT
List competitor accounts for a workspace or specific social account.

list_media_options

ChatGPT
List albums, Pinterest boards, watermarks, and signatures available for your social accounts. Use this before creating posts that need a board, album, watermark, or signature by name.

list_publer_accounts

ChatGPT
SHOWS THE ACCOUNTS CARD to the user. Use this only when the user asked to see their accounts — list, show, browse, which accounts do I have, pick an account. When you need account data to carry out a different request, call lookup_publer_accounts instead so no card is drawn. Showing the card also clears the accounts attached to this chat. Each item has id, name, channel, provider, label, and workspace. For create/schedule/publish, always pass the account "id" from this list — never bare name alone (names collide across networks, e.g. Facebook + Mastodon "PublerTesting"). Show users name/channel/label only; never ask them for ids. THE USER'S TAP IN THIS CARD NEVER REACHES YOU — it is recorded on the server, not in this chat, and nothing in this result changes when they pick. So when they next say "those", "the ones I picked", or name nothing at all, do NOT ask them to select and do NOT list again. Call submit_publer_posts or analyze_recent_performance with no account argument and the server scopes it to whatever they chose.

list_publer_drafts

ChatGPT
List saved drafts: unpublished posts tied to a social account. Ideas — saved without an account — are NOT drafts; list them with list_publer_ideas. Call this when the USER asks for drafts, saved drafts, or draft posts — do not claim the connector is unavailable. It draws a card and starts the post selection over, so never call it to check your own work — a completed result is the confirmation. Each post includes a ref object { n, id } for update/delete/publish_publer_draft. To publish a draft, call publish_publer_draft with that ref after the user confirms.

list_publer_ideas

ChatGPT
List saved ideas: content kept without any social account (created with create_publer_ideas), shared with the workspace or private. Not drafts — a draft belongs to an account; use list_publer_drafts for those. Call this when the USER asks for ideas, saved ideas or content ideas. It draws the posts card on its Ideas filter and starts the post selection over. Each idea includes a ref object { n, id } for update/delete. An idea cannot be published as it is: it has no account to go out on.

list_publer_labels

ChatGPT
The workspace's labels — name and color. Call this before labeling a post so you reuse the user's existing spelling instead of creating a near-duplicate. Pass type to list only labels in use on posts, media or posting-schedule slots. Publer has no way to create a label on its own: it is created the first time a post uses it (labels on submit_publer_posts / update_publer_post).

list_publer_media

ChatGPT
SHOWS THE MEDIA GALLERY CARD to the user. Use this only when the user asked to see their library — show, list, browse, pick media. When you need a media ref to attach to a post you are already building, call lookup_publer_media instead so no card is drawn. Showing the card also clears the files attached to this chat. THE USER'S TAP IN THIS CARD NEVER REACHES YOU — it is recorded on the server, not in this chat, and nothing in this result changes when they pick. So when they next say "those", "the ones I picked", or name nothing at all, do NOT ask them to select and do NOT list again. Call the create/schedule tool with no media argument and the server attaches whatever they chose. Returns markdown images hosted on this MCP. Paste those ![name](url) lines into the chat so the user sees pictures — never names-only. To use an asset on a post, copy ref from the matching item into posts[].media_id. Use search to find assets by name or caption. Use limit to cap how many results are returned. The card shows the folders and opens them when tapped, and loads more files as the user scrolls: NEVER name the folders, count pages or offer to show more files in your reply. Pass folder_id only when the user names a folder to open — the ref.id of that folder in this result's folders[].

list_publer_posts

ChatGPT
List posts with full Publer filter support (state, dates, accounts, search). NEVER call it to find posts the user has already attached — delete_publer_posts and publish_publer_draft take them with no refs, edits take the attached post's caption + account, and listing first draws a second card for one instruction. Default state is published — for drafts prefer list_publer_drafts (or pass state "draft"). It draws a card and starts the post selection over; to act on a post you have no ref for, pass its caption + account to that tool instead. Each post includes a ref object { n, id } for update/delete. Never ask the user for a post id.

list_publer_workspaces

ChatGPT
Read the workspace names. Do NOT call this to ask the user which workspace to use — it draws no picker, so asking from here makes the user read a list instead of tapping one. When a workspace is needed, just call the tool the user actually asked for: it returns the picker card by itself. Use this only when the user explicitly asks which workspaces exist. If only one workspace exists it is already locked — do not ask the user to choose. Never show ids. To switch later, call change_publer_workspace.

load_publer_card_state

ChatGPT
Widget-only. Returns what the card at this turn already committed, so a reopened conversation shows what was done instead of a fresh list. Not for the assistant to call.

lookup_publer_accounts

ChatGPT
Account data with NO card. Use this whenever you need account ids, names or channels to carry out something else the user asked for — analytics, creating, scheduling or publishing a post, or resolving a name the user typed. Never use it when the user asked to see their accounts; that is list_publer_accounts.

lookup_publer_media

ChatGPT
Media data with NO card and no images. Use this when you need a media ref id to attach to a post you are already building, or to check whether a named asset exists. Never use it when the user asked to see their media; that is list_publer_media. Files filed in a folder are not listed until you pass that folder's folder_id; the result's folders[] names the folders at this level (folder_path[] is the trail back to the root). To find a file the user says is in a folder, call again with that folder's ref.id.

lookup_publer_posts

ChatGPT
Widget-only, and it draws no card: the posts card calls this when the user scrolls for more or changes a filter on the card itself. Never call this yourself — use list_publer_posts or list_publer_drafts so the user gets the card.

publish_publer_draft

ChatGPT
Publish an EXISTING draft immediately (preferred way to publish drafts). Pass the ref object { n, id } from a drafts listing, caption + account when the user names the draft in chat, or nothing when exactly one post is attached — the confirmation then names it and returns its ref for the call after the yes. Requires user_confirmed:true after the user explicitly says yes. Does not create a duplicate: publishes the draft content then removes the original draft. Never tell the user you cannot publish a draft — use this tool instead of refusing.

read_publer_docs_page

ChatGPT
Read a Publer API documentation page. Uses the local corpus when available, otherwise fetches live markdown from publer.com/docs.

read_publer_help_article

ChatGPT
Read a Publer Help Center article. Uses the local corpus when available, otherwise fetches live from publer.com/help.

reschedule_publer_post

ChatGPT
Move an existing scheduled post or draft to a new date/time. Use it for a time-only change — it moves the post without rewriting it, and re-creating the post from a copied caption duplicates it. DOES NOT require asking the user for an id: pass the ref from the posts card or the post's caption + account, plus scheduled_at. Example: { "ref": { "n": 0, "id": "..." }, "scheduled_at": "2026-10-14T09:30:00+02:00" }. Keep the UTC offset the post is already shown in. Only the member who created the post can move it, and a post already in the publishing queue cannot be moved at all.

save_publer_card_state

ChatGPT
Widget-only. Records what the card at this turn did, so it survives a reload or leaving the conversation. Not for the assistant to call.

search_publer_docs

ChatGPT
Search the Publer API documentation corpus first, then enrich with live GitBook Q&A if needed. Use before composing advanced post payloads (polls, carousels, link posts, platform formats).

search_publer_help

ChatGPT
Search Publer product help articles (how-to guides, troubleshooting, connecting accounts). Uses the local corpus first, then live help search.

select_publer_account

ChatGPT
Attach social accounts to what the user asks for next (from the accounts card Use button, or chat). Pass account ids (preferred) or labels from list_publer_accounts — accounts[] for several, account for one. The attached set scopes the following requests: create/schedule/publish may omit account and will use every attached account, and analyze_recent_performance covers each; per-account analytics use it only when exactly one account is attached. It holds until the user changes it or clears it. Call it with NO account and NO accounts to read the current set without changing it; pass clear:true to remove it.

select_publer_media

ChatGPT
Record Media Library selection for this conversation (multi-select). Pass refs from list_publer_media. Use replace to set exactly these ids, clear to empty. Call with NO refs to read what is attached without changing it. Then create/schedule with no media argument — attached files are used on their own.

select_publer_posts

ChatGPT
Attach posts to what the user asks for next (from the posts card Use button, or chat). Pass the ref object from list_publer_posts for each one. Deleting takes the attached posts and publishing takes a single attached draft with no refs, so the user can say "delete them" without naming each. Call with NO refs to read the current set without changing it; pass clear:true to remove it. Selecting never acts on a post: destructive tools still ask for their own confirmation.

select_publer_workspace

ChatGPT
Lock this conversation to one Publer workspace, by the name the user gave. Omit workspace only to re-show the list. After this, omit workspace on other tools unless switching.

submit_publer_posts

ChatGPT
The one way to put posts into Publer. when picks what happens to them: - "draft" (default) — saved as a draft. Dated ≠ scheduled: scheduled_for defaults to ~1h ahead so it shows on the Calendar, but the status stays Draft. - "schedule" — goes out at posts[].scheduled_for, which is then required. - "now" — publishes immediately. Only when the user explicitly asked to publish; otherwise draft. To publish an EXISTING draft use publish_publer_draft instead. - "auto" — Publer picks the time inside posts[].range_start/range_end. Needs posting schedule slots on the account; with none the post is NOT created, so use "schedule". - "recycle" — repeats on the cadence in posts[].recycling. - "recur" — repeats on the calendar rule in posts[].recurring. RULES, all modes: pass the account id from lookup_publer_accounts, never a bare name when names collide — or omit account and the accounts the user attached are used: every post without an account goes to EVERY attached account, so the total is posts × attached accounts. All posts in one call share one account set — for a total spread across different accounts, make one call per account. One user request is ONE call with one post and one account unless they asked for more; more than one of either returns confirmation_required and creates nothing until confirm_multiple:true. Never copy a caption out of list_publer_posts / list_publer_drafts into a new submit — that duplicates the old post. Never retry after a timeout: poll get_publer_job_status first, since the same caption+account is blocked for 10 minutes unless allow_duplicate:true. If the user attached or generated a file in this chat, call create_publer_post_from_file or create_publer_photo_draft instead — never put /mnt/data paths in posts[] and never media_base64. For link, poll, carousel, story, reel or document posts, call search_publer_docs first to confirm the payload shape.

update_publer_label

ChatGPT
Rename or recolor an existing label. It changes the label on EVERY post, file and posting-schedule slot that carries it, and a rename also updates the timeslot and recycling rules that used the old name. Needs workspace admin or owner rights. To put a label on one post, use update_publer_post with add_labels instead.

update_publer_post

ChatGPT
Edit an existing Publer post in place — scheduled, draft or published. Changes caption (text), title, media, account, scheduled time and labels. Labels: add_labels / remove_labels change one label and keep the rest; labels replaces the whole set. Pass ONLY what changes, and NAME it: the server reads the post and applies your fields to it, so everything you leave out is kept. A call carrying nothing but a ref changes NOTHING and is refused — it is not a way to apply something the user mentioned earlier. To attach or replace a picture, pass media_id (or media_ids) explicitly, from lookup_publer_media / upload_publer_media. Files the user tapped in the media card are used only when the call names no other change. DOES NOT require asking the user for an id: pass the ref from the posts card already shown, or the post's caption + account. Example: { "ref": { "n": 0, "id": "..." }, "text": "New caption" }. Copy list_filters if the list was filtered. New media replaces the post's media rather than adding to it. For a time-only change prefer reschedule_publer_post — it moves the post without rewriting it. Returns the updated post details from Publer. Published-post edits are network-limited (X/Pinterest/Instagram/Threads/Bluesky only allow label changes after publish).

upload_publer_chat_media

ChatGPT
Upload a ChatGPT-attached or generated image/video into Publer. REQUIRED: top-level file (preferred) or media — ChatGPT injects via openai/fileParams. Prefer create_publer_post_from_file to draft+attach in one step. Never pass /mnt/data paths on other args (connector file-rewrite errors). Returns media_id for posts[].media_id (set post_type to video for .mp4).

upload_publer_media

ChatGPT
Upload an image/video for use in posts. For ChatGPT attachments prefer upload_publer_chat_media. Use top-level media { download_url, file_id } or file_url — never /mnt/data or file_path. Direct uploads return validity metadata so you can pre-check network/post-type compatibility.

upload_publer_media_from_url

ChatGPT
Import media into Publer from public URL(s). Pass url for a single file, or urls[] for bulk. Prefer attaching media_url directly on create/schedule tools when possible. Returns media_id / media_ids and validity metadata when Publer provides them.

Capabilities

WritesInteractive

Example Prompts

Click any prompt to copy it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/publer"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/publer)