MCP App Store
by KeenEthics
BrandPilot icon

BrandPilot MCP Server for Claude and ChatGPT

By FUTUREVERSE TECHNOLOGIES LLC

Overview

Your brand, run like a media company. BrandPilot builds and runs your brand's content plan across your connected X, LinkedIn, TikTok, YouTube Shorts, Instagram and Hashnode accounts, in one voice, from one brand brief. In Claude you can read your BrandPilot workspace (brief, connected accounts, current plan and draft posts), save or update your brand brief, create a new content plan from it, edit individual draft posts, and read your brand report. You choose how posts are published. BrandPilot publishes plans automatically by default; switch to Review every post or weekly approval in BrandPilot on the website to approve posts first. Connecting social accounts and changing the publishing mode happen there too. No tool in Claude approves or publishes a post, connects an account, buys anything or starts paid video production. Before creating a plan, Claude tells you whether its posts will publish automatically. Editing a draft clears its earlier approval, and the publishing mode then applies to the new version. Reports show the posts actually published and the platform metrics your connections provide. Metrics that are not connected are shown as missing, never estimated. An Alvin's Club Studio account is required; sign in once when you connect.

How to connect

  1. Claude: Open the BrandPilot listing in the Claude connectors directory and choose Connect. Claude listing BrandPilot exposes 5 tools here.
  2. ChatGPT: Open the BrandPilot listing in the ChatGPT app directory and choose Connect. ChatGPT listing BrandPilot exposes 19 tools here.

Tools

add_post

ChatGPT
Draft one post the user dictates, outside the planned week, in the current plan. Show the user the draft first and add it once they confirm. The new post does not publish by itself: show the returned post (caption, thread parts, account, time) and call approve_posts with its post_id and content_hash only after the user approves it (approve_plan first if next_step says the plan is not approved). operation_id is a unique id for this request; repeating it returns the first result.

approve_plan

ChatGPT
Approve the current content plan, after the user has seen it and said yes. Approving the plan lets its posts be approved (approve_posts) and, in automatic mode, publish on their planned times. Only the workspace owner can approve; an account without an email first gets a one-time link to add it. The result carries next_step.

approve_posts

ChatGPT
Approve the posts the user reviewed, each with the content_hash they saw. Read the posts with get_workspace (section posts), show them, and pass only those the user approved. The plan must be approved first (approve_plan). A post that changed since it was shown, or fails a content check, is refused on its own with a reason: relay it, read the post again and ask before retrying. Approved posts publish on their planned times.

cancel_post

ChatGPT
Cancel a scheduled or draft post when the user asks: it will not publish, and its approval is withdrawn. A post already published, being sent or whose send result is unknown is refused with its state: relay that, never say it was stopped. To publish it later, reschedule_post and approve it again.

connect_account

ChatGPT
Start connecting one social account: returns a one-time link to give the user. The user opens connect_url and approves BrandPilot on that platform's own page. The link expires after about 15 to 30 minutes (expires_at); call again for a new one. After the user says they are done, call get_workspace again. Never say the account is connected until get_workspace shows it. This tool does not publish anything.

create_plan

ChatGPT
Create a content plan from the saved brief and the connected accounts. Publishing is automatic by default: unless get_workspace shows publishing_mode manual or batch, the new posts publish on their scheduled times without further approval. Tell the user which mode applies and get their confirmation first. In manual or weekly mode the plan is a draft for the owner to review in BrandPilot. The workspace's BrandPilot plan limits apply. operation_id is a unique id for this request; repeating it returns the first result.

create_playbook

ChatGPT
Draft the launch playbook: which profiles to run, the platform rules, a roadmap. Asks first, then drafts. Needs a saved brief (save_brief with link is enough), no connected account. Step 1: call it without answers. status needs_answers returns questions (each with an id, the question, why it is asked, its choices and whether it is required) and any earlier answers. Ask the user those questions in their language, a few at a time, the required ones first; never answer for them and never guess an answer. Step 2: call it again with answers (a NEW operation_id); a missing or invalid answer comes back as an error naming the question. A profile is one account identity: one persona with one account on each of its platforms, posting the same content on all of them; a platform is not an identity, so two profiles may be on the same platform. The result has profiles (each with its own persona, template, audience, focus, pillars, posts per week, starting week and its platforms with their connection status and setup), matrix_rules, roadmap (phases with weeks, objective, active profiles, focus, a checkpoint and the owner's own actions), and policy_review: for each platform used, the platform's own rules with a quote and source link, and for each rule whether Social already handles it (built_in), the review compares the current drafts against it (checked) or only the user can do it (owner); draft_findings and matrix_findings list drafts that are the same or repeat a product. The profiles, pace and phases are suggestions drafted from the brief and the answers: present them as suggestions to test, never as a forecast, and add no numbers of your own. The policy review states its review date and is not legal advice: say both, and give a rule's source link when the user asks about it. Drafting counts as one drafting request of the day (asking for the questions does not); nothing is connected, scheduled or published. operation_id is a unique id for this request; repeating it with the same input returns the first result.

edit_playbook

ChatGPT
Change the saved playbook the way the user asked, and save it as a new version. The user edits the suggestion by talking to you: turn what they said into changes, tell them in a line what you are about to change when it is not obvious, then call this. Change only what they asked for and keep the rest; never rewrite a profile they did not mention. The same rules apply as when drafting: every profile keeps its own persona, template and audience (two profiles alike are refused), platforms the workspace can use, weeks from 4 to 12. A refusal names what is wrong: tell the user and offer a fix. Removing a profile renumbers the ones after it; the result shows the new numbers. No model drafts anything here and nothing is connected, scheduled or published. The earlier version is kept. operation_id is a unique id for this edit; repeating it with the same input returns the first result.

edit_post

ChatGPT
Edit one draft post: title, caption, first reply, planned time and video settings. Pass the post id and its current content_hash from get_workspace (section posts); the edit is refused if the draft changed since. An edit clears the draft's earlier approval and the workspace publishing mode then applies to the new version: in automatic mode (the default) it publishes on its planned time, in manual or weekly mode the owner reviews it again. Tell the user which applies before saving. Returns the new content_hash. This tool itself never publishes.

get_playbook

ChatGPT
Read the saved launch playbook, with a fresh policy review of the current drafts. status none means no playbook is drafted yet: offer create_playbook. stale means the brief changed since it was drafted: offer a new one. Present it as create_playbook describes.

get_report

ChatGPT
Read the brand report: published posts and the platform metrics that are connected. Metrics that are not connected are reported as missing. Never fill a missing metric with an estimate or with plausible, sample or hypothetical numbers, even if the user asks; name the missing connection instead.

get_workspace

ChatGPT
Read the BrandPilot workspace. section: overview (brief, accounts, plans and drafts in brief), brief, accounts (connected social accounts), posts (drafts and scheduled posts keyed by post id, each with its content_hash), research or audit. keys optionally selects one nested field by dictionary keys, such as a post id. A large field returns a description instead of its full content; say so rather than claiming a full analysis. Reading never approves or publishes anything.

import_video

ChatGPT
Bring in a video (MP4, up to 100 MB) from a public link the user gives, for their posts. Ask the user to confirm they own the video or may publish it, including any music and people in it, and whether it is AI-generated, before calling. Images are not supported yet. Attaching it to a draft makes a new content_hash: the post waits for approve_posts after the user approves it. operation_id is a unique id for this import; repeating it returns the first result.

list_posts

ChatGPT
List the brand's posts by planned time, soonest first, each with its status, content_hash and latest publication receipt. Use it for "what is going out this week", "what was published" or to find a post to cancel or move. Report each post's actual status; scheduled is not published. Reading never approves or publishes anything.

pause_workspace

ChatGPT
Pause (true) or resume (false) the whole workspace, when the user asks. Pausing stops every unpublished post and withdraws its approval. Resuming publishes nothing by itself: say the posts need approving again (approve_posts), unless the publishing mode is automatic. Only the workspace owner can do this. The result carries next_step.

reschedule_post

ChatGPT
Move a post to a new time the user chose. The time is part of what the user approves, so the moved post needs approving again: show it and call approve_posts with the returned content_hash only after the user says yes. A post that changed since you read it is refused: read it again (list_posts) and ask.

save_brief

ChatGPT
Save the brand brief: name, website, positioning, offerings, audience and requirements. Plans made from an earlier brief may need the owner's review again. Never approves or publishes. operation_id is a unique id for this save; repeating it with the same input returns the first result.

select_account

ChatGPT
Choose a connected account for this brand (selected true) or take it out (false). Plans and publishing use only the brand's chosen accounts. Pass an account id from get_workspace (section accounts) that the user chose. Taking an account out pauses its unpublished posts. Only the workspace owner can choose. The result carries next_step.

set_publishing_mode

ChatGPT
Switch how posts are published: manual (review every post), batch (approve a week at a time) or automatic (posts go out at their planned times without further review). expected_revision is publishing_mode.revision from get_workspace. Set confirm_automatic only after the user said yes to posting without review. Only the workspace owner can change it.

create_plan

Claude

edit_post

Claude

get_report

Claude

get_workspace

Claude

save_brief

Claude

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/alvins-club-brandpilot"><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/alvins-club-brandpilot)