MCP App Store
by KeenEthics
Petit Panda icon

Petit Panda MCP Server for ChatGPT

By Petit Panda

Overview

Petit Panda is a social media content tool: create images and videos, then publish or schedule them on your social accounts, from ChatGPT. Generate images and videos with the models of the Petit Panda Studio, save them as drafts, then publish them now or schedule them on the accounts you connected to Petit Panda: Instagram, TikTok, YouTube, LinkedIn, Facebook, Threads and X. Before scheduling, Petit Panda checks images and videos against each network's documented format rules and tells you what to fix. You can also read your post analytics, reply to comments and to Instagram or Messenger DMs, and manage your comment and DM automations. Publishing, sending a message, deleting and starting a generation are confirmed with you first. A Petit Panda account is required: you sign in once when you connect, and the plugin only accesses your own workspace.

How to connect

  1. ChatGPT: Open the Petit Panda listing in the ChatGPT app directory and choose Connect. ChatGPT listing Petit Panda exposes 77 tools here.

Tools

cancel_generation

ChatGPT
Stops a queued or running generation. Credits come back only when the provider confirms the cancel: the result says refunded. Ids come from list_generations or generate_*. Always confirm with the user before calling this tool.

create_automation

ChatGPT
Creates an Instagram or Facebook automation that automatically sends a DM or a link when a trigger fires (keyword in a comment or DM, any comment, story reply, welcome DM). It is created paused unless status is explicitly active, and an active one starts messaging real people. Returns { automation }. Always confirm with the user before calling this tool.

create_automation_template

ChatGPT
Saves a reusable automation template. Optionally makes it the default template of one account, which replaces that account's previous default. Returns { template }.

create_character

ChatGPT
Creates a reusable character from 1 to 4 photos. Free and ready right away (no training). Then pass its id as character_id to generate_image / generate_video: its photos are used as references by the models that accept images.

create_draft

ChatGPT
Creates a draft post (nothing is published or scheduled). Media ids come from import_media or the uploads tools. Returns { draftId }. Schedule it afterwards with create_post.

create_post

ChatGPT
Creates a post on a connected social account and publishes it publicly, now or later. Publishes immediately when scheduledTime is absent; schedules it when scheduledTime (ISO 8601) is given or useNextFreeSlot is true. mediaUrls = array of public image/video URLs; [] for a text-only post. Call list_accounts first to get accountId. Media and text are checked against the platform's rules BEFORE anything is created: an invalid format returns a detailed error instead of a post that would fail silently. Use media_specs if you produce the media yourself. Always confirm the final text, account and timing with the user before calling this tool: the post becomes public on their account. Use the user's text exactly as they wrote it: never add hashtags they did not ask for. YouTube: never add #Shorts to the title or description; a vertical video of 3 minutes or less is classified as a Short automatically.

create_schedule_slot

ChatGPT
Adds a recurring weekly posting slot. A duplicate slot is refused with 409. Returns { slotId }.

create_support_ticket

ChatGPT
Opens a numbered support ticket with the Petit Panda support team, who answer by email. Returns { ticketNumber, subject }. Always confirm with the user before calling this tool.

create_upload_url

ChatGPT
Hosts a LOCAL file (image or video) on Petit Panda storage and returns its public URL, to pass in mediaUrls of create_post. Use it ONLY when you can run an HTTP PUT yourself (a terminal or code environment). In a chat client that cannot make HTTP requests, do NOT use it: call import_media instead (with the files attached by the user, or with public https URLs). Not for media that is already a publicly reachable URL. Returns { uploadUrl, publicUrl }. Steps: 1) call this tool with the filename; 2) upload the raw bytes with an HTTP PUT to uploadUrl; 3) put publicUrl in the mediaUrls array of create_post. For a carousel, repeat for each image and keep the order. Check the target format with media_specs first: a wrongly sized image is accepted here but REJECTED by the platform at publishing time. Creates a private storage slot only; nothing is published.

delete_automation

ChatGPT
Permanently deletes an automation. Its run history goes with it. Returns { ok: true }. Always confirm with the user before calling this tool.

delete_automation_template

ChatGPT
Permanently deletes a saved template. Automations already created from it are not affected. Returns { ok: true }. Always confirm with the user before calling this tool.

delete_character

ChatGPT
Deletes a reusable character (ids from list_characters). Past generations are kept. Always confirm with the user before calling this tool.

delete_comment

ChatGPT
Removes a comment thread (id from list_comments) from the user's Petit Panda comment inbox only: the comment stays on the social network. Returns { ok }. Always confirm with the user before calling this tool.

delete_draft

ChatGPT
Permanently deletes a draft. Returns { ok: true }. Always confirm with the user before calling this tool.

delete_generation

ChatGPT
Removes a generation from the Studio library (cancels it first if still running; credits come back only if the provider confirms that cancel). Media already published are not affected. Always confirm with the user before calling this tool.

delete_schedule

ChatGPT
Cancels a scheduled post so it is never published. Call list_schedules first for the scheduleId. Only 'pending' posts can be cancelled. The post moves to status 'cancelled' and leaves the default lists; its draft and media are kept intact. Already published posts are never touched.

delete_schedule_slot

ChatGPT
Deletes a recurring posting slot (already scheduled posts are not affected). Returns { ok: true }. Always confirm with the user before calling this tool.

disconnect_account

ChatGPT
Disconnects (revokes) a connected social account so it can no longer be published to; it must be reconnected from the web app. Prefer accountId (from list_accounts) to disconnect a single account; passing only platform revokes EVERY account of that platform in the workspace. Provide at least one of them. Returns { success: true }. Always confirm with the user before calling this tool.

estimate_cost

ChatGPT
Gives the EXACT cost in credits of a generation before it starts (same price as the Studio and as the real charge), the current balance and the balance after. Same parameters as generate_image / generate_video (model, aspect_ratio, resolution, duration, count, params). Charges nothing. Call it before each generate_image / generate_video to tell the user the cost. Read-only.

generate

ChatGPT
Starts an image, video or avatar generation in the Petit Panda Studio. Asynchronous: returns a generation id right away (status queued or running), to follow with get_generation until succeeded, failed or refused. Uses the user's credits at the same rate as the web app (see list_models), refunded automatically on a refusal or a failure. Once succeeded, pass the output URL (outputs[0].url) in mediaUrls of create_post to publish it. BEFORE calling this tool: call estimate_cost with the same parameters, tell the user the exact cost in credits (the total when count > 1) and their remaining balance, and wait for their explicit OK. After the generation, tell them the credits used (costCredits) and the balance left (balance).

generate_image

ChatGPT
Generates one or more images with a Petit Panda Studio model. Uses credits at the price shown by list_models (refunded automatically on a failure or a refusal; a prompt refused by the Petit Panda safety filter is never charged). Waits for the result by default (wait=true); wait=false returns the id right away, to follow with get_generation_status. Returns id, status, costCredits, balance (credits left), the result url(s) once 'succeeded', and next = the ready-to-use create_post call to publish or schedule the result. BEFORE calling this tool: call estimate_cost with the same parameters, tell the user the exact cost in credits (the total when count > 1) and their remaining balance, and wait for their explicit OK. After the generation, tell them the credits used (costCredits) and the balance left (balance).

generate_video

ChatGPT
Generates one or more videos (or a talking avatar) with a Petit Panda Studio model. Many video models start from an image: pass start_image (a URL or the id of an image generated before). Uses credits at the price shown by list_models (refunded on a failure or a refusal). Returns the id right away by default (wait=false), to follow with get_generation_status; wait=true waits up to about 100 s, then returns the id if the video is not ready. Same response as generate_image, including next to publish with create_post. BEFORE calling this tool: call estimate_cost with the same parameters, tell the user the exact cost in credits (the total when count > 1) and their remaining balance, and wait for their explicit OK. After the generation, tell them the credits used (costCredits) and the balance left (balance).

get_account_analytics

ChatGPT
Detail for one connected account: its daily metrics series over a period and its tracked posts sorted by impressions (up to 200). The id comes from list_accounts. Returns { period, account, series, posts, total, demo }, or 404 if the account is not the user's. Read-only.

get_account_quota

ChatGPT
Lists the workspace's connected social accounts WITH the plan's account quota (used, per-platform limit and usage), and can include inactive, expired or revoked accounts with their status, which list_accounts hides. Use it to see why an account cannot be used or whether another can be connected. Returns { accounts: [{ id, platform, account_handle, account_name, status, is_active }], quota: { plan, planLabel, used, perPlatformLimit, perPlatformUsed } }. Read-only.

get_accounts_analytics

ChatGPT
Daily account metrics (followers, follower growth, profile views, reach, impressions) grouped by connected account over a period, plus each account's latest snapshot. Filter by platform or by account id (from list_accounts). Returns { period, accounts, demo }. Read-only.

get_analytics_overview

ChatGPT
Aggregated KPIs over a period across all connected accounts: reach, impressions, follower growth, number of posts and average engagement rate, with a per-platform breakdown. A metric a platform does not provide is null, not 0. Returns { period, kpis, byPlatform, demo }. Read-only.

get_analytics_post

ChatGPT
Detail of one tracked post: all its raw metrics, a comparison to the average of the other posts on the same account, and whether it is the account's best post. The id comes from list_analytics_posts. Returns { post, demo }, or 404. Read-only.

get_automation_stats

ChatGPT
Per-automation statistics (sends, reads, link clicks) over a period, plus a daily series. Returns { period, automations, daily }. Read-only.

get_character_training

ChatGPT
Returns the Soul ID training status of a character, refreshing it while the training runs (the stored status is updated, nothing else changes).

get_conversation

ChatGPT
Returns one DM conversation (id from list_conversations) with its message history and whether a reply can still be sent (Instagram and Messenger allow replies for 24 hours after the contact's last message, or 7 days for a reply written by a person). Returns { conversation, messages }. Read-only.

get_credits

ChatGPT
The workspace's credit balance, its plan and the date the monthly credits renew. Check it before calling a tool that uses credits. Returns { balance, plan, resetAt }. Read-only.

get_draft

ChatGPT
Reads one draft: title, caption and first comment per platform, attached automations, media ids, hashtags, language, status and dates. Get draft ids from list_drafts. Returns { draft }. Read-only.

get_generation

ChatGPT
Returns the status and result of a generation started with generate: progress, status (queued, running, succeeded, failed, refused, canceled), outputs (URL of the media hosted by Petit Panda once succeeded), and the error and refund when it failed. Read-only.

get_generation_status

ChatGPT
Status of a generation started by generate_image / generate_video (or generate): status (queued, running, succeeded, failed, refused, canceled), progress 0-100, costCredits, refunded, the result url(s) once 'succeeded', the error otherwise. wait=true waits for the end (about 100 s max). Once 'succeeded', next gives the ready-to-use create_post call. Read-only.

get_post

ChatGPT
Returns the full detail of a scheduled or published post from its scheduleId: platform, account, title, full text, number of media, status, dates, URL and error if any. Use list_schedules to get the scheduleId. Read-only.

get_post_remake_origin

ChatGPT
Tells whether a tracked post was generated in Petit Panda Studio and returns the prefill (prompt, model, settings) needed to remake it. The id comes from list_analytics_posts. Returns { prefill }, or 404 when the post has no Studio origin or is not the user's. Read-only, nothing is generated.

get_today_activity

ChatGPT
Real signals about the user's publishing activity: posted yesterday, current streak of consecutive days with a published post, pending posts scheduled for today, drafts waiting, and posts published in the last 7 days. Returns { postedYesterday, streakDays, scheduledToday, draftsWaiting, publishedThisWeek }. Read-only.

get_upload

ChatGPT
Returns the result of a link created by request_upload. Status 'pending' means the user has not finished sending yet (ask them to finish, then call again, do not poll in a tight loop). Status 'ready' returns urls: the public URLs of the uploaded files, in the order the user picked them, to pass in mediaUrls of create_post. Read-only.

get_user

ChatGPT
Returns the authenticated Petit Panda profile: plan, number of connected social accounts and their platforms. Read-only. Use it to check the connection works before any other call.

import_media

ChatGPT
Imports images or videos into Petit Panda storage and returns their public URLs, in order, to pass in mediaUrls of create_post. Use it when the user attached files in the chat (passed in files) or when you have public https URLs (urls). Accepts jpg, png, webp, gif, mp4, mov, webm (images up to 20 MB, videos up to 60 MB, max 10 files per call); the type is checked on the content-type AND the file bytes. Returns { media: [{ publicUrl, kind, mimeType, size }], publicUrls }. For a carousel, import all the images in one call and keep the order. Check the target format with media_specs first. Stores the files only; nothing is published.

inspect_media

ChatGPT
Reads one or more media files from public https URLs, WITHOUT downloading them whole (HTTP Range on the header), and returns for each: type (image or video), MIME type, file size, width, height, duration (video), aspect ratio, container and codec. For each platform requested (default: all of them) it returns status 'ok', 'warning', 'blocked' or 'unverified' (a file could not be read) plus the list of problems found against that platform's documented rules: severity 'error' means the platform will refuse the file, 'warning' means it may be refused, cropped or re-encoded. Use it on a file BEFORE create_post, or to check a file you were handed. Width and height are the DISPLAY size (a phone video stored 1920x1080 with a 90 degree rotation is reported 1080x1920). A field the file does not expose, such as the duration of a fragmented MP4, comes back null and is listed in 'unverified': never read 'ok' as fully verified then. Rules the platform documentation does not state are listed in 'notDocumented' (unknown, not allowed-by-default). For the complete rules table with documentation links, call media_specs. YouTube is judged as a Short unless youtubeFormat is 'long'. Only public https URLs are read (private or local addresses are refused). Read-only: nothing is stored or published.

invite_setter

ChatGPT
Invites a person by email as a setter on one of the user's Instagram accounts, so they can handle its DM inbox. Sends a real invitation email. Returns { grant, acceptUrl, emailed }. Always confirm with the user before calling this tool.

list_accounts

ChatGPT
Lists the social accounts connected to Petit Panda (TikTok, LinkedIn, Instagram, Facebook, Threads, YouTube, X). Returns each account's id, platform and handle. Call it BEFORE create_post to get the right accountId. Optional platform filter. Read-only.

list_analytics_posts

ChatGPT
Lists the user's published posts with their metrics (impressions, reach, likes, comments, shares, saves, video views, engagement rate). With period, only posts published in that window. Each post has a composite id to pass to get_analytics_post. Returns { posts, total, limit, offset, sortBy, order, demo }. Read-only.

list_automation_runs

ChatGPT
Lists the 50 most recent runs of one automation (what triggered it, status sent, failed or skipped, error message, time). Returns { runs }. Read-only.

list_automation_templates

ChatGPT
Lists the user's saved automation templates (favorites), optionally for one platform. A template can be the default for one account. Returns { templates }.

list_automations

ChatGPT
Lists the user's comment and DM automations with their trigger, action, status (active or paused) and send stats (sent, failed, skipped). Returns { automations }.

list_characters

ChatGPT
Lists the user's characters (id, name, photos, creation date), newest first. Read-only.

list_comments

ChatGPT
Lists comment threads left on the user's posts (and on accounts shared with them as a setter), newest first, with a public permalink per post. Returns { comments, permalinks, pagination }. Comment ids returned here are the ones used by get/update/delete/reply tools.

list_conversations

ChatGPT
Lists the user's DM conversations (and those of accounts shared with them as a setter) for a channel, most recent first, up to 200. Each has contact info, status, segment and the messaging window state. Returns { conversations }. Conversation ids are used by get_conversation, send_dm and update_conversation.

list_drafts

ChatGPT
Lists the drafts of the active workspace, most recently updated first. Returns { drafts: [{ id, title, content_per_platform, first_comment_per_platform, pending_automation_per_platform, media_ids, hashtags, language, status, created_at, updated_at }] }.

list_generations

ChatGPT
Lists the user's Studio generations, newest first: id, kind, model, status, outputs (hosted URLs), cost in credits, favorite. Also returns the current credit balance. Use before (the created_at of the last item) to page. Read-only.

list_models

ChatGPT
Lists the Petit Panda Studio models (image, video, avatar) that the generation tools can use: id, type, accepted parameters (aspect ratio, duration, resolution...), expected inputs and the cost in credits of each parameter combination. Call it before generating to pick a valid model and parameters. Read-only.

list_notifications

ChatGPT
Lists the 30 latest notifications of the user's workspace (publication results, failures, etc.) and the count of unread ones. Returns { notifications: [{ id, type, title, body, href, read_at, created_at }], unreadCount }. Read-only.

list_schedule_slots

ChatGPT
Lists the recurring weekly posting slots (day, time, timezone, platform, optional account, auto_fill, is_active). Returns { slots }.

list_schedules

ChatGPT
Lists the user's posts (scheduled, published, failed), filterable by status and platform over a date window, ordered by publication date. Returns each post's account, title and a text excerpt (300 characters max); use get_post for the full detail. For a 'failed' post, the error_msg field explains exactly why the platform refused it and what to fix: report it to the user instead of only saying it failed. If the error is about image dimensions, call media_specs, regenerate the visuals at that size, then create the post again. A 'published' post WITHOUT platform_post_url can be an older post from before link tracking existed. Read-only.

list_setters

ChatGPT
Lists the setter invitations the user created (a setter is a teammate who answers DMs on one of the user's Instagram accounts without access to the account itself), with the invited email and status. Returns { grants }. Read-only.

list_team_members

ChatGPT
Lists the members of the active workspace (name and role) and its seats. Returns { org: { name, seatCount, plan, includedSeats, maxSeats }, role, members: [{ fullName, role, isSelf }] }. Read-only: members are invited or removed from the web app.

mark_notifications_read

ChatGPT
Marks one notification (by id, from list_notifications) or all unread notifications as read. Provide either id or all: true. Returns { ok: true }.

media_specs

ChatGPT
Returns each platform's documented constraints: accepted image dimensions and aspect ratios, file formats, max file size, max number of media, and text limits (caption, title, alt text) with how they are counted. It also returns the VIDEO rules of each platform under video (duration, aspect ratio, resolution, file size, containers, codecs), every value with its documentation URL and quote, plus what the documentation does not state. To check a real file against these rules, call inspect_media. Call it BEFORE generating or uploading media, so the media is produced in the right format instead of being rejected at publishing time. Key rules: TikTok photo = 1080x1920 (9:16), 1080p maximum, a larger image is REJECTED; Instagram ratio strictly between 4:5 and 1.91:1; Threads ratio up to 10:1. Never guess these values, ask for them. Read-only.

next_free_slot

ChatGPT
Returns the next free posting slot for a platform, based on the recurring slots the user configured in Petit Panda (looks 14 days ahead, avoids collisions). Pass the returned scheduledAt as scheduledTime to create_post. Read-only.

replace_draft_media_and_comments

ChatGPT
Edits an existing draft through the full-replace path: media_ids, hashtags, first_comment_per_platform, content_per_platform and pending_automation_per_platform each REPLACE the stored value when passed, and stay untouched when omitted. Returns { draftId }.

reply_to_comment

ChatGPT
Publicly replies to a comment (id from list_comments) on Instagram, Facebook or Threads, then marks the thread as replied. Other platforms return a not-supported error. Returns { ok, platform_comment_id }. The reply is visible to everyone on the network. Always confirm with the user before calling this tool.

request_upload

ChatGPT
Use this when the user wants to add a local photo or video (from their phone or computer) and you cannot send the file bytes yourself (a chat client without file upload or HTTP access). Returns a short link (valid 30 minutes, tied to the user's account). Give the link to the user, ask them to open it, pick one or several files and wait for the confirmation on the page. Then call get_upload with the returned token to receive the public URLs, in the order the files were picked, and pass them in mediaUrls of create_post. Creates nothing public by itself; nothing is published.

revoke_setter

ChatGPT
Revokes a setter invitation: the setter loses access to the inbox immediately. Returns { ok: true }. Always confirm with the user before calling this tool.

send_dm

ChatGPT
Sends a text DM reply in a conversation (id from list_conversations). Allowed within 24 hours of the contact's last message, or up to 7 days as a reply written by a person (marked automatically); beyond that it fails with window_closed. Returns { ok, message }. The contact receives it on Instagram or Messenger and it cannot be unsent. Always confirm with the user before calling this tool.

set_generation_favorite

ChatGPT
Adds a generation to the favorites (favorite: true) or removes it (false). Ids come from list_generations.

sync_comments

ChatGPT
Pulls the latest Threads replies and Facebook Page comments into the comment inbox and resubscribes the Pages to the comment webhook. Throttled server-side, so it is cheap to call. Returns a sync summary. Only works on the user's own connected accounts.

train_character

ChatGPT
Starts a Soul ID training on a character so its face is reproduced more faithfully by the models that support it. Returns the character and the remaining balance (credits refunded if the training service is down). Follow it with get_character_training. Uses 375 credits. BEFORE calling this tool, tell the user it uses 375 credits (get_credits gives their balance) and wait for their explicit OK.

update_automation

ChatGPT
Edits an automation: send only the fields to change. Setting status to active makes it start messaging real people, paused stops it. Returns { automation }. Always confirm with the user before calling this tool.

update_automation_template

ChatGPT
Edits a saved template: send only the fields to change. Platform and trigger/action types cannot be changed. Returns { template }.

update_comment_status

ChatGPT
Sets the status of a comment thread (id from list_comments) to unread, read, replied or hidden. On Facebook, hidden also hides the comment on the Page (and any other status unhides it); the status only changes if that platform call succeeds. Returns { ok, status }. Always confirm with the user before calling this tool.

update_conversation

ChatGPT
Sets the status of a DM conversation (id from list_conversations) to open, closed or archived. Does not message the contact. Returns { ok, status }.

update_draft

ChatGPT
Edits a draft: only the fields you pass change, and content_per_platform is fully replaced (send every platform you want to keep). To change media or first comments use replace_draft_media_and_comments. Returns { ok: true }.

update_schedule

ChatGPT
Edits a post that is scheduled but not yet published (status 'pending' only): its text, title, date/time and/or platform options. Call list_schedules first for the scheduleId. Give text to rewrite the caption, title for the title (YouTube/TikTok), scheduledTime (future ISO 8601) to move it, and/or options to change a setting afterwards (e.g. Instagram { trialReel: true }, TikTok { privacyLevel: 'SELF_ONLY' }). At least one of the four. text REPLACES the current caption; the options given are merged into the existing ones.

update_schedule_slot

ChatGPT
Changes a posting slot; only the fields you pass change. Slot ids come from list_schedule_slots. Pass socialAccountId null to unbind the account. Returns { ok: true }.

upload_widget

ChatGPT
Shows an interactive 'Drop a photo or video' zone inside the chat so the user can pick one or several local images or videos (jpg, png, webp, gif, mp4, mov, webm; up to 10 files) from their device. The files are uploaded directly to Petit Panda storage, then the widget posts the resulting public URLs, in the order the user chose, as a new user message. Use it when the user wants to post media that exists only on their device and you cannot upload bytes yourself. After calling it, wait for the user's message containing the URLs, then pass them in mediaUrls of create_post. The tool itself uploads nothing; its only argument is the optional maxFiles.

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/petit-panda"><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/petit-panda)