Use DeepSeek to research TikTok and Instagram trends
Updated
Virlo is social listening for TikTok and Instagram Reels. DeepSeek models reason and write well, but they know nothing about the TikTok and Instagram videos breaking out this week. Mount Virlo in DeepSeek Harness, or call it next to the DeepSeek API, and DeepSeek works from live trends, hooks, sounds and outlier videos drawn from a database of 23.1M viral videos. DeepSeek is an AI lab whose models power the DeepSeek chat app, the DeepSeek API and DeepSeek Harness (dsh), an open-source agent harness where everything is a plugin.
- One YAML row in DeepSeek Harness
- Tools appear as mcp__virlo__ in every session
- A plain REST route for DeepSeek API code
Part of Virlo's AI assistant guides. Running a model locally instead? See Ollama. Want every MCP option in one place? See the MCP server guide.
How do I connect Virlo to DeepSeek Harness?
DeepSeek Harness talks to MCP servers through its own client plugin, @deepseek-ai/dsh-mcp-client. You add one row for Virlo to your profile's patch file and restart.
Step 1: Create your Virlo account for free
Step 2: Start DeepSeek Harness once
Launch dsh from npm with Node installed. It opens the web UI at
127.0.0.1:3080and creates yourwebprofile, which is where the Virlo row will live.Terminalnpx @deepseek-ai/dsh webStep 3: Export your Virlo key
Generate a key on the API Keys page (it starts with
virlo_tkn_) and expose it to the shell that runs dsh, so the YAML can read it instead of storing it in plain text.Terminalexport VIRLO_API_KEY=virlo_tkn_YOUR_KEYStep 4: Add the Virlo row to cordis.patch.yml
Merge this insert patch into
$DSH_HOME/profiles/web/cordis.patch.yml(by default under~/.dsh). Merge rather than overwrite, since the file may already hold your other patches. Then stop and restart dsh.cordis.patch.yml- insert: - id: mcp-virlo name: '@deepseek-ai/dsh-mcp-client' config: serverName: virlo transport: streamable-http url: https://dev.virlo.ai/api/mcp/mcp headers: Authorization: !!js '`Bearer ${process.env.VIRLO_API_KEY}`'Step 5: Ask DeepSeek
Open a new session in the web UI. The Virlo tools appear in the model's tool list as
mcp__virlo__followed by the tool name.Your first promptUse Virlo to list this week's 5 TikTok trends to copy in [your niche]. For each, say what to keep, what to change for [your brand], and give one example video.
Want the row in every profile on this machine? Put the same patch in $DSH_HOME/cordis.patch.yml instead of the per-profile file.
DeepSeek chat vs DeepSeek API vs DeepSeek Harness: how does each reach Virlo?
| DeepSeek surface | What it is | How Virlo connects | Best for |
|---|---|---|---|
| DeepSeek Harness (dsh) | Open-source agent harness built on Cordis plugins | A dsh-mcp-client row with transport: streamable-http | Research sessions where DeepSeek calls Virlo tools itself |
| DeepSeek API | OpenAI-compatible endpoint at api.deepseek.com | Your code fetches Virlo's REST API, then sends results to the model or exposes them as a function tool | Apps, scripts and scheduled jobs |
| DeepSeek models in an MCP client | A DeepSeek model selected inside Cline, OpenCode, Goose or similar | The client's own MCP config pointed at the Virlo server | Coding agents that already use DeepSeek as the model |
| DeepSeek chat app | The consumer chat on web and mobile | Paste results from one of the routes above | Quick rewrites of research you already pulled |
The model is the same everywhere. What changes is whether DeepSeek can call Virlo on its own or needs your code to fetch first.
What does DeepSeek see when it asks Virlo for trends?
The five trends to copy with the most creators posting in the last 7 days, each with a real example video: the data behind an mcp__virlo__ trends call in DeepSeek Harness.
In Virlo's data, 659 creators posted "My moon my man couple montage" in the last 7 days, the most of any trend to copy as of November 19, 2026.
As of November 19, 2026 at 13:07 UTC

1. My moon my man couple montage
Fresh · 659 creators this week

2. While they're out partying btw
Rising · 346 creators this week

3. He wanna pay full price SNL trend
Fresh · 175 creators this week

4. But you're not a Buddhist trend
Fresh · 158 creators this week

5. Shoulder carry mirror selfie trend
Fresh · 158 creators this week
| Trend | Stage | Creators this week | Creators last week |
|---|---|---|---|
| My moon my man couple montage | Fresh | 659 | new this week |
| While they're out partying btw | Rising | 346 | 115 |
| He wanna pay full price SNL trend | Fresh | 175 | new this week |
| But you're not a Buddhist trend | Fresh | 158 | 33 |
| Shoulder carry mirror selfie trend | Fresh | 158 | new this week |
See all 39 trends to copy
Recipes, 30-day charts, client fit and every example video are in the Virlo app.
Open Trends to copyHow this sample is made
- Source
- Virlo's Trends to copy: formats found by Virlo's trend detector and published only after the team checks each one.
- Window
- Creators who posted each trend in the last 7 days, counted from TikTok and Instagram videos only. Trends made with an AI tool are left out.
- Method
- Ranked by creators posting this week. Stages: Fresh, Rising, Peaking, Fading.
- As of
- November 19, 2026 at 13:07 UTC. Refreshes every hour.
- Made by
- Numbers are pulled automatically from Virlo's data every hour; the text around them is written by the Virlo team.
Suggested citation: Source: Virlo, “Use DeepSeek to research TikTok and Instagram trends”, virlo.ai/deepseek
How do I use the DeepSeek API with Virlo data?
The DeepSeek API has no MCP client of its own, so your code does the fetching. Call Virlo's REST API at https://api.virlo.ai/v1 with your key, then either paste the JSON into the prompt or declare a function tool that wraps the call, so the DeepSeek model decides when to fetch.
- Read:
curl -s "https://api.virlo.ai/v1/content-trends?platform=tiktok&limit=5" -H "Authorization: Bearer $VIRLO_API_KEY" - Send the response to your DeepSeek model with a system prompt such as "write three scripts that keep each trend's structure for [your brand]".
- For tool calling, describe a
get_trendsfunction withplatformandlimitparameters and run the Virlo request when the model calls it. - For alerts, register a webhook with
POST /v1/webhookson events liketracking.outlier_video.detected, and hand each payload to DeepSeek for a short write-up.
Full endpoint and webhook references live in the Virlo API docs.
What can I ask DeepSeek once Virlo is connected?
Paste into a DeepSeek Harness session with the Virlo row loaded. Replace the brackets.
Reasoned trend pick
Pull this week's TikTok trends to copy in [your niche], then reason through which two best fit [your brand] and why.
Hook taxonomy
Collect 20 hooks from [your niche] videos with a Virality Score of 18 or more and group them by hook type with a one-line pattern for each.
Script rebuild
Fetch the transcript of [TikTok link], break it into hook, setup, payoff and call to action, and rebuild it for [your product].
Competitor teardown
Study @[competitor] on TikTok and Instagram. Which videos beat their usual views, and what pattern links them?
Sound shortlist
Which Instagram Reels sounds and TikTok sounds are breaking out this week? Recommend one of each for [campaign].
Data to JSON
Return the top 15 TikTok videos about [topic] from the last 30 days as JSON with views, Virality Score and hook text.
Weekly plan
Draft a five-day posting plan for [your brand] from live Virlo trends, with the source video for each idea.
DeepSeek chat vs DeepSeek API vs DeepSeek Harness: which guide is this?
DeepSeek is a model family with several front doors. This page covers all three and the route each one takes.
DeepSeek Harness (this page, setup above)
The dsh agent harness with its MCP client plugin. Virlo goes in as one YAML row.
DeepSeek API (this page, section above)
Pay-per-call model access for developers. Pair it with Virlo's REST API in your own code.
Kimi
A different model maker with its own chat and coding tools. Its guide covers that setup.
Local models
Running a DeepSeek model on your own machine through Ollama? Connect Virlo in the client you chat through.
What are the limits?
- Virlo reads public TikTok and Instagram data. DeepSeek can analyse and write with it, but no route on this page posts to social accounts.
- DeepSeek Harness reads patch rows at boot, so restart dsh after editing
cordis.patch.yml. If the first connection fails, dsh still starts and logs the error. - The dsh stdio bridge strips credential-looking variables from child processes; the remote Virlo row reads its key from
process.envinside the header instead. - Trend data refreshes every hour after a Virlo team review.
- See what each plan includes on pricing.
Frequently asked questions
Does DeepSeek support MCP?
DeepSeek Harness does. It ships an MCP client plugin, @deepseek-ai/dsh-mcp-client, that connects to one server per row over stdio or Streamable HTTP. The DeepSeek API itself has no MCP client, so code using the API fetches Virlo through REST or function tools.
How do I add an MCP server to DeepSeek Harness?
Add an insert patch with a row named @deepseek-ai/dsh-mcp-client to $DSH_HOME/profiles/web/cordis.patch.yml. Give it a serverName, transport, url and, for remote servers, headers. Restart dsh and the tools appear as mcp__ plus the server name plus the tool. If the server cannot be reached at boot, dsh still starts and writes the error to its log.
What is DeepSeek Harness?
DeepSeek Harness, run as dsh, is DeepSeek's open-source agent harness. It is built on Cordis, so tools, providers and connections are plugins mounted into a profile. Start it with npx @deepseek-ai/dsh web and it serves a web UI on your machine.
Can I use the DeepSeek API for TikTok research?
Yes, with a data source. The model has no live TikTok feed, so fetch from Virlo's REST API, for example GET /v1/content-trends?platform=tiktok, and pass the results to DeepSeek. A function tool lets the model request the data itself when it needs it.
Where is cordis.patch.yml?
Under your dsh home, which defaults to ~/.dsh. A profile patch sits at profiles/[profile]/cordis.patch.yml, for example profiles/web/cordis.patch.yml, and a machine-wide patch sits at cordis.patch.yml in the home folder. Merge new rows in rather than replacing the file, because it may already hold patches you or other plugins added earlier.
How do I keep my Virlo key out of the DeepSeek Harness config?
Export VIRLO_API_KEY in the shell that starts dsh and reference it in the row's headers with a !!js process.env expression, as in the setup above. The key then never sits in the YAML, and rotating it only means changing the variable.
Can DeepSeek watch TikTok videos?
Not directly. Through Virlo it receives what is inside each video: the spoken hook, on-screen text, format, view counts, Virality Score and, on TikTok, the full transcript. That lets DeepSeek reason about why a video took off and write a new one on the same structure.
Can DeepSeek post to social media through Virlo?
No. Virlo only reads public TikTok and Instagram data and has no access to your accounts. DeepSeek can research, rank and draft from it inside a Harness session or your own API code, and you publish the finished post with whatever scheduling or posting tool you already rely on.
Give DeepSeek live social media data
One row in DeepSeek Harness and the model can check what is working before it writes.
More AI assistant guides
- AI assistant guides
- Claude
- Claude Code
- Cursor
- MCP server
- TikTok MCP server
- Instagram MCP server
- ChatGPT
- ChatGPT dots
- OpenAI Codex
- Gemini CLI
- Grok Build
- Claude Cowork
- n8n
- Zapier
- Grok Bot
- Muse
- Claude Duet
- Hermes Agent
- Kimi K3
- Cue
- Manus
- Perplexity Computer
- Paperclip
- OpenClaw
- NanoClaw
- AI agents CLI
- ChatGPT agent
- Perplexity
- Microsoft Copilot
- Le Chat
- Notion
- Slack
- Discord
- Telegram
- Google Sheets
- Windsurf
- VS Code
- GitHub Copilot
- Cline
- JetBrains
- Kiro
- Google Antigravity
- goose
- Warp
- OpenCode
- Zed
- Amp
- Factory Droid
- Raycast
- Lovable
- Replit
- Bolt.new
- v0
- Ollama
- LM Studio
- Open WebUI
- LibreChat
- Make
- Pipedream
- Gumloop
- Lindy
- Relevance AI
- Airtable
- ClickUp
- Dify
- Flowise
- LangChain
- CrewAI
- OpenAI Agents SDK
- Mastra
- Vercel AI SDK
- Claude Agent SDK