- Brand
- Cospark
- Category
- Marketing
- Primary Subcategory
- Multi-Modal Generative Media Platforms
Integration details
Description
Cospark helps marketers and creators research references and produce ad media from ChatGPT and Codex. Search public social content from Instagram, TikTok, YouTube, Meta Ad Library, Reddit, and X; explore Cospark's ad library; generate images, video clips, talking-head UGC ads, and voiceovers; and inspect or transcribe media. Organize approved brand details, research, scripts, and assets in Cospark project files, or create a workspace to continue editing in the Cospark app. Connect your Cospark account through OAuth. Generation and social research can consume Cospark credits. Supplied media may be sent to processing providers. Generated and imported media and inspection images are available at public asset URLs; do not upload confidential media. Project file tools can replace, move, or delete files when requested. With a separately connected TikTok account, upload one video or up to 35 photos as a draft for review and completion in TikTok. Media is sent to TikTok; this does not publish a public post. The plugin cannot launch ad campaigns, manage ad spend, or edit workspace timelines through MCP. This plugin connects directly to the hosted Cospark MCP. Workflow guidance is retrieved with read_docs; no bundled skills are required.
- Integration type
- Plugin
- Verification status
- Not applicable
- Platform
- ChatGPT
- Primary Subcategory
- Multi-Modal Generative Media Platforms
- Secondary Subcategories
- None listed
- Brand
- Cospark
- Access
- Account required
- First tracked
- 2026-11-21
- Tool count
- 133
- Market
- US
The Primary Subcategory used for this profile’s headline score.
Other Subcategories where the Integration is listed.
Not discoverable in ChatGPT yet
ChatGPT now suggests Plugins on its own when they match a user's request, but not every Plugin is being surfaced yet.We haven’t seen Cospark suggested yet. Sign up and we’ll track it, and your Discoverability Score will appear here once ChatGPT starts surfacing it.
What discovery looks like

Get alerts for Cospark
Get updates when Cospark’s Discoverability Score or category rank changes.
Competing in ChatGPT Multi-Modal Generative Media Platforms
View Category133 tools agents can invoke
Check a UGC video run started by generate_ugc_video. Poll the same run ID until status is completed or failed; never restart generation merely because it is still processing. A completed result can be partial: disclose failedScenes and partial rather than claiming every scene succeeded.
get_ugc_video_status
Check runs started by generate_video, generate_video_from_frames, or generate_video_with_references. Pass runIds to check several runs in one call. Poll until each status is completed or failed; never restart generation merely because it is still processing. Failed runs are not charged; error carries the provider's reason. A run rejected as an invalid request is not retried, so change the prompt or input frames before starting a new one.
get_video_generation_status
Create an empty Cospark ad workspace with chat, editor, canvas, and storyboard that the user can open and edit in the Cospark app. Returns the session ID and a URL to open it.
create_workspace
Create an original realistic adult character from a short description, without scraping or copying a real creator. Generates one silent 5-second 9:16 Seedance 2.5 clip at 480p by default. Consumes credits and returns a run ID; poll get_character_status to extract its native opening frame and save the reusable person in your character library. Generated media URLs are public. Starting another call spends credits on another character.
create_character
Create a Cospark project, such as one per app, product, or brand. Each project has its own file system for the brand details, guidelines, research, scripts, and media used to make its ads and marketing content.
create_project
Create a folder, including missing parents, in an authorized Cospark project's access-controlled file tree. Creates only project metadata within Cospark, without publishing content or creating external folders. Succeeds without changes if the folder already exists.
create_project_folder
Delete a file or folder from a Cospark project. A folder that is not empty is only deleted with recursive set to true. Media that was uploaded or imported into this project from a URL is deleted from Cospark too once no project entry uses it; generated or library media stays in the user's Cospark library.
delete_project_file
Edit an existing text file in an authorized Cospark project without rewriting it. Changes are confined to Cospark's access-controlled project storage; this does not publish text or edit external files. Each edit either replaces exact text (oldString must match once, or set replaceAll) or inserts text at the start or end of the file. Edits apply atomically. Returns the new version and the line where each edit landed; read the file again to see the result.
edit_project_file
Retrieve public Facebook content: comment replies. Get the replies to a comment. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_post_comment_replies
Retrieve public Facebook content: comments. Fetches comments from a Facebook post or reel with cursor-based pagination. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_post_comments
Retrieve public Facebook content: event details. Get a specific event by its URL or id This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_event_details
Retrieve public Facebook content: events. Get the events of a city. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_events
Retrieve public Facebook content: facebook group info. Fetches the public information shown on a Facebook group's About page, including its description, privacy and visibility, member and activity counts, categories, administrators and moderators when Facebook exposes them, group history, and rules. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_group
Retrieve public Facebook content: facebook group posts. Fetches posts from a public Facebook group, limited to 3 posts per page due to API limitations. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_group_posts
Retrieve public Facebook content: marketplace item. Fetches details for a Facebook Marketplace item by item id or Marketplace item URL, including title, description, price, location, condition, photos, seller, and availability flags. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_marketplace_item
Retrieve public Facebook content: marketplace location search. Searches Facebook Marketplace locations/cities and returns coordinates you can use with the Marketplace Search endpoint. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_marketplace_location_search
Retrieve public Facebook content: marketplace search. Searches Facebook Marketplace listings by keyword and lat/lng. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_marketplace_search
Retrieve public Facebook content: post. Retrieves a single public Facebook post or reel by URL. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_post
Retrieve public Facebook content: profile. Retrieves public Facebook page details including category, address, email, phone, website, services, priceRange, rating, likeCount, talkingAboutCount, and followerCount. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_profile
Retrieve public Facebook content: profile events. Get the events of a public Facebook page This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_profile_events
Retrieve public Facebook content: profile photos. Fetches photos from a public Facebook page with pagination support. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_profile_photos
Retrieve public Facebook content: profile posts. Returns publicly visible Facebook profile posts, limited to 3 posts per page due to API limitations. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_profile_posts
Retrieve public Facebook content: profile reels. Fetches up to 10 reels per request from a public Facebook page. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_profile_reels
Retrieve public Facebook content: search events. Search for events by name. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_events_search
Retrieve public Facebook content: transcript. Extracts the transcript text from a Facebook video post or reel. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_facebook_post_transcript
Create one complete 9:16 talking-head UGC video ad from an exact script, creative direction, and one or more ordered starting frames. A single frame is reused for every scene; multiple frames are assigned to scenes in order without generating additional variations. Use this for a full ad body, not a single replacement shot. Generated media is stored at publicly accessible URLs. This consumes credits; start it once, then poll the same run ID with get_ugc_video_status until the reviewed final composition is ready.
generate_ugc_video
Create or edit an image asset for an ad. GPT Image 2.5 Sunburst is the default; Nano Banana 2.1 (gemini-nano-banana-2.1), Nano Banana 2 (Gemini 3.1 Flash Image), and Nano Banana Pro (Gemini 3 Pro Image) are optional model overrides. Optionally provide one or more ordered reference images to preserve or combine products, people, characters, composition, or style. Pass background: transparent with GPT Image 2.5 Sunburst for a PNG cutout with an alpha channel. Reference inputs may be Cospark file IDs or public URLs. Generated media is stored at publicly accessible URLs. This consumes credits and returns a reusable Cospark file ID and URL.
generate_image
Create a standalone video ad shot from text. Generated media is stored at publicly accessible URLs. This consumes credits and returns a run ID immediately; poll that same ID with get_video_generation_status until it completes or fails, and never restart a run that is still processing.
generate_video
Create a video ad shot from an exact starting image and optional ending image. Use it for UGC hooks, product motion, or a targeted replacement clip. Generated media is stored at publicly accessible URLs. This consumes credits; poll the returned run ID with get_video_generation_status. Inputs may be Cospark file IDs or public URLs.
generate_video_from_frames
Create a video ad shot guided by reference images, video, or audio that should shape the result without being exact first or last frames. Use images for identity, product, or setting, a reference video for motion or when editing a supplied video, and audio for voice or sound; say in the prompt which parts of each reference to use. Do not use a reference video to reproduce another creator's video or likeness. Generated media is stored at publicly accessible URLs. This consumes credits and returns a run ID immediately; poll that same ID with get_video_generation_status. Inputs may be Cospark file IDs or public URLs.
generate_video_with_references
Create voiceover audio for a video ad, including duration and timestamped transcript segments. Generated media is stored at publicly accessible URLs. This consumes credits; use list_voices first when the user has not chosen a voice.
generate_voiceover
Fetch a compact evidence document for one ad by ID. The document is returned as readable text, media as native MCP resources, and only the stable ad ID as structured output. Image documents include brand, placements, CTA, dimensions, and first/last seen dates. Indexed video documents include the persisted summary, hook, CTA, audio treatment, timestamped transcript, and approximate semantic scenes with visible text. Use after search_ads for analysis or comparison; video scene boundaries are not verified frame-accurate cuts.
get_ad
Resume create_character by run ID. Once the video completes, extracts its first frame as a PNG and saves it in your character library. Returns characterId, reusable image fileId/url and the source video. Repeated polls reuse the saved frame and character. If extraction_failed, poll this same ID to retry extraction without generating or charging for another video. Inspect the returned image before using it; extraction does not assess visual quality. Run handles expire after seven days; saved characters and files persist.
get_character_status
Inspect a video or audio asset directly. Videos return a labeled contact sheet sampled at 0.2s, 1.2s, 2.2s and so on, capped at 64 frames spanning the full video. Frames are 160 pixels wide with an automatic grid. Includes frame timestamps, duration, and a timestamped transcript if audio is present. Audio-only sources return transcription without images. Read the sheet to understand visual content. Prefer it over downloading media and running your own frame extraction or transcription. Use it for generated, uploaded, or external media. For an Instagram, TikTok, or Meta Ad Library post, pass the post link itself; Cospark fetches the video. When only the spoken content matters, the platform transcript tools are enough; for an indexed ad-library video, use get_ad, which already returns its stored analysis. Inputs are limited to 100 MiB and 10 minutes: compress a large local file with ffmpeg before uploading it, since inspection doesn't need delivery quality. For denser or larger frames, build a contact sheet locally; the Cospark tools guide has the commands.
inspect_media
Retrieve public Instagram content: basic profile. Fetches a lightweight Instagram profile summary by user ID, returning username, full name, biography, profile picture URL, verification status, follower count, following count, media count, and account privacy and type. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_instagram_basic_profile
Retrieve public Instagram content: comment replies. Retrieves the public replies to a specific Instagram comment. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_instagram_post_comment_replies
Retrieve public Instagram content: comments. Retrieves comments on a public Instagram post or reel. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v2_instagram_post_comments
Retrieve public Instagram content: embed html. Returns the raw HTML embed snippet for an Instagram user's profile widget. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_instagram_user_embed
Retrieve public Instagram content: get reels by audio id. Fetches the reels Instagram exposes for an audio page like instagram.com/reels/audio/{audio_id}/. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_instagram_audio_reels
Retrieve public Instagram content: highlights details. Fetches the full contents of a specific Instagram story highlight album by its ID. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_instagram_user_highlight_detail
Retrieve public Instagram content: popular search. Use this to explore an Instagram topic and the posts Instagram curates for it. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_instagram_search_popular
Retrieve public Instagram content: post/reel info. Fetches detailed metadata for a single Instagram post or reel by shortcode or URL. This tool does not publish posts, send messages, follow accounts, or modify the source social content. If download_media=true, the provider copies media into permanent public storage and returns the new URLs; this creates stored assets and costs additional credits. Only request that copy when the user asks to save media. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_instagram_post
Retrieve public Instagram content: posts. Returns a paginated feed of a user's public Instagram posts, including reels, photos, videos, and carousels. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v2_instagram_user_posts
Retrieve public Instagram content: profile. Retrieves comprehensive public Instagram profile information including biography, bio links, follower and following counts, verification status, and profile picture URLs. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_instagram_profile
Retrieve public Instagram content: profile post count. Returns the post count shown on a public Instagram profile. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_instagram_profile_post_count
Retrieve public Instagram content: reels. Returns a paginated list of a user's public Instagram reels (short-form videos). This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_instagram_user_reels
Retrieve public Instagram content: search hashtag posts. Use this when you know the exact hashtag and want Google-indexed public Instagram posts or reels, optional date filters, and pagination. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_instagram_search_hashtag
Retrieve public Instagram content: search instagram. Use this for Instagram-native account, hashtag, or place lookup. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_instagram_search
Retrieve public Instagram content: search instagram profiles. Use this for Instagram-native profile lookup. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v1_instagram_search_profiles
Retrieve public Instagram content: search reels. Use this when you only want Google-indexed Instagram reels matching a keyword or phrase, with optional date filters and pagination. This tool does not publish posts, send messages, follow accounts, or modify the source social content. This call consumes Cospark credits based on the ScrapeCreators credits actually charged. Returns compact agent-friendly JSON by default; use responseMode="raw" only when complete upstream fields are necessary.
v2_instagram_reels_search
Cospark ChatGPT Plugin FAQ
How the directory, categories and Discoverability Score work.
Read the methodologyHow do I improve Cospark's ChatGPT Plugin discoverability?
The levers are the listing surface agents actually read: names, descriptions, keywords, tool metadata, and registry health. Which lever matters depends on where discovery breaks, which is what continuous measurement shows.
What are Cospark alternatives on ChatGPT?
As of 2026-11-21, Cospark competes with 1TapEdit AI Studio, Abracadabrax Images & Video, Ad Army, Animx, Arcads, ArivuStudio, Artlist, Atlas Cloud and 59 more in ChatGPT Multi-Modal Generative Media Platforms, ranked by public Discoverability Score.
Where is this profile measured?
This profile uses the geography attached to the latest public registry snapshot: US. Locale tags are intentionally omitted.