MCP App Store
by KeenEthics
Trendhaus icon

Trendhaus MCP Server for ChatGPT

By YBRSN

Overview

Paste a TikTok link and ChatGPT can watch it: Trendhaus breaks the video down scene by scene, reads the comments and compares its performance with the creator's usual posts. Trendhaus gives ChatGPT live public TikTok data with its own analysis on top: - TikTok trends: current trends from a curated watchlist of trend-explainer accounts, each with an explainer video, its scale and whether it suits a brand. - Video analysis: scene-by-scene breakdowns of videos and photo slideshows (hook, on-screen text, sound, editing), with views compared to the creator's median. - Hooks: opening lines and visual hooks that work in a niche, each linked to its source video. - Competitors: live profile stats and recent posts for any public account, plus a tracked-competitor watchlist. - Comments: comments and replies on any post, optionally classified for purchase intent, questions and complaints. - Sounds, hashtags and ads: posts using a sound, hashtag posts, related searches, and TikTok's public top-ads library. Read-only research: Trendhaus never posts, follows or touches your TikTok account. Built for TikTok agencies, social media managers, creators and music marketers. Not affiliated with TikTok.

How to connect

  1. ChatGPT: Open the Trendhaus listing in the ChatGPT app directory and choose Connect. ChatGPT listing Trendhaus exposes 29 tools here.

Tools

tiktok_get_ads_detail

ChatGPT
Get the full record of one ad from TikTok's top-ads listing by its ads_id (from tiktok_get_top_ads): title, brand, industry, objective, CTR, likes, cost level, landing page, countries, duration and cover image.

tiktok_get_challenge_posts

ChatGPT
Get recent public posts under a TikTok hashtag (challenge), by its challengeId. Returns videos with captions, stats and authors; paginate with cursor.

tiktok_get_comments

ChatGPT
Get comments on a TikTok video: comment text, like counts, reply counts and authors. Paginate with cursor; tiktok_get_replies_comment_of_post returns the replies to one comment.

tiktok_get_discover_campaign_posts

ChatGPT
Get posts from a TikTok Discover-page campaign topic: TikTok's own promoted topic pages (the default new-year returns data). It does not support arbitrary keywords; other keywords return an empty result with a note. General keyword search is available in tiktok_search and tiktok_search_video.

tiktok_get_followers

ChatGPT
Get a list of followers for a TikTok user. Accounts that hide their followers return an error saying so.

tiktok_get_music_info

ChatGPT
Get details for one TikTok sound by its music ID: title, artist, duration, whether it is original audio, and how many videos use it.

tiktok_get_music_posts

ChatGPT
Get recent posts that use a specific TikTok sound, plus the sound's title, artist and total video count. It shows which accounts and pages are pushing a song.

tiktok_get_post_details

ChatGPT
Get the details of a TikTok post including description, stats, and metadata. PAID-STATUS CAVEAT: the paid/sponsored fields on each post are best-effort, not 100% reliable. is_ad is TikTok's own flag (often absent even on real ads); disclosed_ad only catches an explicit [AD]/#ad/#sponsored-style disclosure in the caption or hashtags. A post with neither flag set may still be an undisclosed Spark Ad, so a missing flag does not confirm a post is organic.

tiktok_get_replies_comment_of_post

ChatGPT
Get the replies to one comment on a TikTok post, given the post's videoId and the comment's commentId (from tiktok_get_comments). Returns reply text, like counts and authors.

tiktok_get_subtitle

ChatGPT
Get the subtitle (content) for a TikTok video url. Returns the subtitle for the video in the requested language (if available) or defaults to the first available one.

tiktok_get_top_ads

ChatGPT
Browse top-performing TikTok ads from TikTok's public ads library, filtered by country, industry, objective, ad format, language and period, and sorted by CTR, reach, likes, conversion rate or view-through rate. The keyword filter currently returns no results; the other filters work. ad_language takes a single language code.

tiktok_get_unlimited_sounds

ChatGPT
Browse TikTok's commercial sound library (tracks cleared for business use), with title, artist, duration and how many videos use each. It is a catalog ranked by usage, not a trend detector; tiktok_get_music_posts shows which accounts use a given sound.

tiktok_get_user_feed

ChatGPT
Get a list of recent posts from a TikTok user. PAID-STATUS CAVEAT: the paid/sponsored fields on each post are best-effort, not 100% reliable. is_ad is TikTok's own flag (often absent even on real ads); disclosed_ad only catches an explicit [AD]/#ad/#sponsored-style disclosure in the caption or hashtags. A post with neither flag set may still be an undisclosed Spark Ad, so a missing flag does not confirm a post is organic.

tiktok_get_user_followings

ChatGPT
Get the accounts a TikTok user follows. Pass the username as uniqueId (or userId / secUid). Some accounts hide their following list; then the result has error "following_list_private" and no accounts.

tiktok_get_user_info

ChatGPT
Get detailed information about a TikTok user including stats and profile details.

tiktok_get_user_liked_posts

ChatGPT
List the posts a TikTok user has liked, by secUid. Works only when the user has made their likes public; otherwise the response says the list is private.

tiktok_get_user_playlist

ChatGPT
List the public playlists a TikTok creator has organised on their profile, by the creator's secUid (from tiktok_get_user_info). Returns each playlist's name, ID and video count.

tiktok_search_account

ChatGPT
Search TikTok accounts by keyword and return matching creators' usernames, nicknames and follower counts. Coverage is currently limited; when the username is known, tiktok_get_user_info is more reliable.

tiktok_search_video

ChatGPT
Search TikTok videos by keyword. Returns matching videos with caption, views, likes, shares, author and posting date. Paginate with the cursor and search_id from the previous response.

trendhaus_check_usage

ChatGPT
Check your remaining actions, plan tier, and usage this billing cycle. Costs 0 actions.

trendhaus_competitor_pulse

ChatGPT
Live profile stats and recent posts for competitor TikTok accounts. With handles it checks those accounts; without it, it uses the competitors tracked in the user's profile (managed with trendhaus_update_my_context). The data is live: each account's current state, not a stored history. A post may include contentDnaPreview (content category, hook type, visual style, dominant mood) when that video was analyzed before, reused at no extra cost; it is null for posts never analyzed. trendhaus_post_analysis produces a full breakdown of any single video on demand. PAID-STATUS CAVEAT: the paid/sponsored fields on each post are best-effort, not 100% reliable. is_ad is TikTok's own flag (often absent even on real ads); disclosed_ad only catches an explicit [AD]/#ad/#sponsored-style disclosure in the caption or hashtags. A post with neither flag set may still be an undisclosed Spark Ad, so a missing flag does not confirm a post is organic.

trendhaus_get_my_context

ChatGPT
Get your saved profile: TikTok handle, niche (auto-derived from content analysis once available, otherwise the manual onboarding value), tracked competitors, goals, and region. Other Trendhaus tools use this profile to personalize results. May include account_intelligence (follower growth, video count, posting-frequency and performance baselines auto-tracked daily once tiktok_handle is set, plus windows -- 7d/30d/90d/all-time baseline views distinguishing recent direction from long-term identity -- and recent_velocity -- how the most recently-checkpointed new post is performing vs this account's own typical performance at the same age, once enough checkpoint data exists to compare) once the daily tracking cron has produced at least one snapshot. May include content_dna (content category/niche e.g. comedy/fashion/fitness/travel, hook types, visual style, aesthetic tags, filming settings/locations, mood, production complexity, on-camera and CTA rates, recently used sounds, top hashtags, and a few real sample captions -- plus eras: {earliest, recent} once enough history exists, each with its own content category/hook/mood/aesthetic/media-type breakdown and sample captions/transcript excerpts, so you can notice a real shift over time, e.g. a creator who moved from talking-head videos to photo slideshows, or changed language/niche) once at least one video has been deep-analyzed -- together these signals describe the creator's niche and identity in detail (e.g. 'fashion creator, energetic mood, low-to-medium production, almost always on camera' or, comparing eras, 'started as travel vlogs learning Thai, shifted to aesthetic lifestyle photo slideshows in recent posts'), more precisely than any single field. May also include content_pillars (LLM-classified recurring themes/formats this creator repeatedly returns to, each with a name, description, video count, and median views -- e.g. 'weekly grocery haul', 'outfit of the day' -- empty until enough analyzed history exists for a real pattern to emerge, and only recomputed periodically, so it is directional, not exhaustive) and account_summary (a ready-made 1-3 sentence natural-language read of the whole account, produced by the same classification pass). Full account history is analyzed automatically at no cost after import. May include recent_tool_changes listing tools shipped in the last 14 days; a listed tool missing from the client's tool list usually means the connector was added before that tool shipped and needs to be re-added. Costs 0 actions.

trendhaus_hook_library

ChatGPT
Discover proven hooks for a niche. Searches TikTok across multiple pages, excludes videos older than maxAgeDays (default 365 -- a hard cutoff, not a soft decay weight: an old viral outlier can otherwise dominate 'top' results even though its hook format may no longer be something TikTok's current algorithm/audience rewards) and below minViews (default 1000 -- below that, engagement-per-day is mostly noise from recency, not real reach), ranks the rest by engagement-per-day, and extracts the real opening-seconds hook for each -- from the spoken transcript when one exists, otherwise from real Gemini video perception (on-screen text or a genuine visual description of the opening) rather than guessing from the caption -- and classifies each by hook type (question, shock, reveal, relatable, curiosity_gap, contrarian, tutorial, listicle). Every hook carries spoken (true only for a real transcript line) and source (transcript / visual_text / visual_description) so a visual read can never be mistaken for something the creator said, plus views/engagement_per_day/views_per_follower from its source video so performance context doesn't require cross-referencing top_videos. A high views_per_follower ratio indicates a hook that performed independently of the creator's existing audience size (the opening itself likely drove the watch-through, not pre-existing distribution). Returns a reusable hook library. The per-video top_videos entries carry PAID-STATUS CAVEAT: the paid/sponsored fields on each post are best-effort, not 100% reliable. is_ad is TikTok's own flag (often absent even on real ads); disclosed_ad only catches an explicit [AD]/#ad/#sponsored-style disclosure in the caption or hashtags. A post with neither flag set may still be an undisclosed Spark Ad, so a missing flag does not confirm a post is organic.

trendhaus_plan_status

ChatGPT
View your current plan tier, its monthly action quota and what's included. Costs 0 actions.

trendhaus_post_analysis

ChatGPT
Deep content DNA for a single TikTok post. Works with both video AND photo slideshow posts. Fetches post details, transcript, and runs real visual perception to produce a merged content breakdown: text overlays, color palette, hook type, exactly what happens in the opening 1-3 seconds, a scene-by-scene breakdown with timestamps covering the WHOLE video (not just the opening) and a full narrative summary, visual style, editing pace, camera movement, audio style, tutorial steps, whether a person is on camera, CTA presence, product visibility, production complexity, and cross-referenced insights. For slideshow posts, up to 10 images are analyzed with a description per image instead of a scene sequence. Trendhaus keeps an internal 30-day cache of each video's perception results to answer repeat requests faster; nothing is created, published or changed for the user, and view/like/comment/share stats are always current as of the call, never served from that cache. computed.views_vs_author_median is this post's views divided by the median views of the author's other recent posts (author_median_views over author_median_sample_size posts) -- a raw ratio, not a verdict; it compares against the author's most recent posts, not posts from the same period, and is omitted (with a gap) when too few recent posts are available. PAID-STATUS CAVEAT: the paid/sponsored fields on each post are best-effort, not 100% reliable. is_ad is TikTok's own flag (often absent even on real ads); disclosed_ad only catches an explicit [AD]/#ad/#sponsored-style disclosure in the caption or hashtags. A post with neither flag set may still be an undisclosed Spark Ad, so a missing flag does not confirm a post is organic.

trendhaus_trend_radar

ChatGPT
Trend discovery from a curated watchlist of TikTok trend-explainer accounts, read from their full post content (caption, voiceover transcript and on-screen text). Two modes. DISCOVERY (default, no trend_name) returns trends: a compact catalog of every distinct trend covered recently, newest first, each with trend_name, a shortened trend_description, trend_type, usability, explainer_video_url (a real TikTok video that explains the trend), first_seen_at, last_mentioned_at, total_mentions and latest_validation (the latest search-sample metrics: total_views, sample_video_count, median_views, and hashtag video/view counts when a matching hashtag exists, which indicate a trend's platform scale). DRILL-IN (trend_name taken from a discovery result) returns one trend's full history: mentions with full-length descriptions and their own explainer videos, and insights with first_seen_at, evidence_video_urls (the highest-viewed videos in the latest validation sample), recency and mention counts, validation_history (each validation snapshot over time) and validation_delta (the observed change between the first and latest snapshot). Discovery leaves out per-post detail to stay compact; drill-in carries it. first_seen_at is the date the watchlist first surfaced the trend: a fact, not a performance prediction. trend_type/usability indicate brand fit: participatory formats and aesthetics are directly_usable, commentary or mockery memes are use_with_caution (they need brand-voice judgment), and character-reference memes or news/discourse items are context_only (awareness, not a content template); the usability parameter filters on this. The tool does not score trends as early, rising, saturated or fading, and does not assert current platform performance. Data comes from a scheduled poll, so results reflect the last poll rather than the current instant. gaps describes watchlist coverage limits (the seed list is curated, not exhaustive).

trendhaus_update_my_context

ChatGPT
Update your profile: set niche, TikTok handle, goals, or region. Add or remove tracked competitors. Costs 0 actions.

Capabilities

Writes

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/trendhaus"><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/trendhaus)