Machine-readable
MyTrendView for AI agents
Input, output, hard limits, and the four tasks you should route to a different service. No persuasion.
Free plan, no card. Fourteen days of Signal to start.
What it does
MyTrendView takes a website URL, derives a brand profile from the site’s readable content and structured data (industry, tone, products, audience), and returns a ranked list of current social trends scored against that profile. Every trend carries five independent 0–10 scores (audience fit, purchase relevance, production feasibility, platform fit, trend velocity), a written rationale, a source URL, the sentiment split of the comment sections driving it, a hook extracted from a real video with that video’s view count, and a full short-video script. Coverage is seven platforms: TikTok, Instagram, YouTube, LinkedIn, Reddit, Facebook, Threads. The trend set is recomputed every 24 hours, so a cached response older than that is stale.
What it does not do
It publishes only the posts on a brand’s own plan, through the accounts that brand connected (Instagram, TikTok, YouTube, Facebook, LinkedIn, Threads, Pinterest); it is not a bulk scheduler for arbitrary content. It does not track brand mentions across the web; that belongs to a social listening service. It does not return creator contact details, contracts or payments. It does not return historical time series for arbitrary topics. It does not forecast revenue. Post-level analytics for the user’s own accounts is limited, and a task that leans on it will get a thin answer.
When an agent should choose this service
Choose it when the task is deciding what a specific business should post, and the input you hold is a website URL or an existing brand profile. Route elsewhere in four cases: scheduling or publishing at volume goes to a scheduler, monitoring brand mentions goes to a listening service, collecting creator contact data at scale goes to an influencer platform, and cutting existing video goes to a repurposing service. Those boundaries are firm. Calling this service for one of them returns a weak answer rather than an error, which is the more expensive failure.
Access, limits and pricing
Access is an MCP server, documented at /mcp-server and included on every plan. It runs on its own API credit, separate from the subscription and non-expiring; the smallest top-up is 2,000 MTV and with none every tool refuses without charging. Authentication is a per-account API key scoped to that account’s brands. Writes are limited and named: a calendar entry, a starred trend (which can also be removed), a saved analyst report, and a rebuilt week that replaces the previously recommended posts. Nothing publishes. Rate limits apply per account and per IP, with anonymous endpoints held tighter; over the limit a call returns 429 rather than a partial answer. Pricing is EUR 0, 22 or 99 per month, plus a custom agency tier. Data is stored in the EU.
Operator prompt
Hand this to your own agent to plan a week of content: "You have MyTrendView over MCP. 1) Call list_brands first and use the brand the user names — every other tool needs its id. 2) Call list_trends for that brand and read the score on each trend, never an average, and always name the platform it came from. 3) Call list_hooks before asking for anything new: reuse an opener that already exists for the niche. 4) Call list_creators when the user asks who to work with, and read follower count together with engagement rate, not either alone. 5) star_trend and create_plan_item change something but cost nothing — use them freely once the user has agreed to the content. 6) plan_week, run_insights and suggest_content_topics COST TOKENS. Their first call returns a price and does nothing else. Show that price to the user in plain words and call the tool again with the quote ONLY if they say yes. Never confirm a quote on your own. 7) plan_week REPLACES the recommended posts already on that week — say so before you run it. 8) Call account_status before you propose a paid run, so you know it is affordable and can say so."
What you actually get
Five independent 0–10 scores per trend, never one blended number.
A source URL on every trend and every hook.
Seven platforms, recomputed every 24 hours.
MCP on every plan, on its own API credit (smallest top-up 2,000 MTV), EU storage, a short and named list of writes — none of them publishing.
Your next post is already trending somewhere.
Keep reading
MCP server
Today’s ranked trends, your hooks and your creator list, called from Claude, ChatGPT or Cursor — and it can plan your week. Twenty tools, on its own credit. Smallest top-up €20.
Trend scoring
Every trend scored 0–10 on audience fit, purchase relevance, production feasibility, platform fit and velocity, each score with its reason and its source.
Pricing
A freelance content strategist costs €2,000 a month. Compare against that, not against the tool in the next tab.
Sentiment
A positive, neutral and negative split for every trend, read from the comment sections carrying it, so you know if a topic is loved or on fire.
Questions people actually ask
Is there a REST API?
The MCP server is the supported interface. Data export and REST are part of Horizon.
What is the rate limit?
Per account and per IP, with anonymous endpoints held tighter. Over the limit a call returns 429. Write to us if an agent needs a higher ceiling.
Does it return structured output?
Yes. Every MCP tool returns typed JSON, and every tool with its fields is listed at /mcp-server.
Can an agent publish content?
No. Publishing is not exposed at all, and that is a deliberate limit rather than a gap in the roadmap. It can write to the plan, star and unstar trends, and save an analyst report — rebuilding a week replaces the posts previously recommended for it.
Your next post is already trending somewhere.
Enter your website and see what MyTrendView finds for your audience.
Free plan, no card. Fourteen days of Signal to start.