Search, transcribe and extract structured data from YouTube, TikTok, Instagram, X and six more platforms. One API for your code, one MCP server for your agents, built for one video or a million.
Same API and same response shape in every major language
per hour of video
<30 s
Speech-to-text when a video has no captions
Three ways in
Plug video into your agent, your code or your workflow
Same engine, same credits, same results. Pick the door that fits how you build.
For AI agents
MCP server and Claude plugin
Install the Claude Code plugin in one command, or connect Claude Desktop, Cursor, n8n or any MCP client in two minutes. Your agent can then find, watch and pull data from videos on its own.
terminal
# Claude Code: MCP server + 2 skillsclaude plugin install vidnavigator \ --marketplace \ vidnavigator/vidnavigator-mcp-starter \ --config api_key=$VIDNAVIGATOR_API_KEY
Claude Code plugin with two skills: watch and video-dataset
Four building blocks that work alone or chained together, on every supported platform.
YouTubeTikTokLIVEMy files
skincare routine
most likedthis monthcountry: frmin 100K views
5 skincare mistakes to stop making2.4M views · 310K likes
My honest 30-day retinol results1.1M views · 142K likes
Dermatologist reacts to viral routines860K views · 97K likes
The 3-step routine under 20 €420K views · 51K likes
128 videos · 5 pages mergedresults.json
01 / Search
Find the videos that matter
Search YouTube with AI ranking, TikTok by keyword with sort, date and engagement filters in more than 130 countries, and LIVE rooms broadcasting right now. Pull every video of a profile, or search inside your own files.
YouTube, TikTok, TikTok LIVE, profiles and your own files
Filter by likes, views, publish date and country
Every page fetched, merged and deduplicated for you
0:04First thing: the battery really lasts two days.
0:09The camera is where it falls short.
0:13For 299 dollars, I would still buy it.
No captions found · speech-to-text47 s
02 / Transcribe
Read any video, in any language
Get the platform's own captions when they exist, or accurate speech-to-text when they do not. The response has the same shape for all 10 platforms, with a timestamp on every segment.
Captions or speech-to-text, in 99+ languages
No length limit with async jobs and webhooks
Instagram carousels, X posts and Snapchat highlights included
Describe the fields you need, such as product, price, claims or sentiment, and get validated JSON back for every video. No regex, no prompt engineering, no clean-up step.
Strings, numbers, enums, arrays and nested objects
Output validated against your schema
Instagram, TikTok and other videos without captions are transcribed for you
Prices stay flat this year; a cheaper tier launches in the second quarter.
12:4127:0541:18
Supports“We are not raising prices in 2026”
Contradicts“Enterprise plans go up in March”
Every answer points to the second it was said.
04 / Analyze and verify
Answers that come with proof
Summaries, people, places and answers to your questions for any video. Turn a post on X into a checkable claim, including its attached video, then find the videos that support or contradict it.
Summaries and Q&A on any video or uploaded file
Tweet claim analysis, text and attached video
Semantic search across your own audio and video files
Point VidNavigator at a search, a profile or a list of links. It finds the videos, reads them, and hands back exactly the data you asked for.
TikTok searchYouTubeInstagramX postsYour files
VidNavigator
transcribe
analyze
extract
Typed JSONYour datasetWebhookAI agent
TikTok searchYouTubeInstagramX postsYour files
VidNavigator
transcribe
analyze
extract
Typed JSONYour datasetWebhookAI agent
01
Find
Start from a keyword, a TikTok profile, a single link or your own list of URLs.
02
Read
We fetch the captions or transcribe the audio. No scrapers, proxies or media formats for you to maintain.
03
Understand
AI turns each transcript into a summary, an answer, or the exact fields of your schema.
04
Deliver
Read the result as JSON, receive a webhook when a job finishes, or let your agent use it directly.
A 30-second clip
Synchronous call, answer in the same request.
A three-hour stream
Async job with no length limit. Poll it or wait for the webhook.
Thousands of videos
Run jobs side by side and collect each result as it finishes.
For developers
From API key to dataset in one file
Typed SDKs, a predictable REST API and an OpenAPI spec. Every example below was run against the live API: add your key and it works.
Search TikTok, extract the same fields from every video, save a CSV
The whole pipeline in about thirty lines: one search, one extraction job per video running in parallel, one row per result.
import csvfrom vidnavigator import VidNavigatorClient, VidNavigatorErrorclient = VidNavigatorClient() # reads VIDNAVIGATOR_API_KEY# 1. Find this month's most-liked TikToks for a keywordsearch = client.search_tiktok( query="creatine gummies", sort_by="most_liked", published_within="this_month", country="gb", max_results=20,)videos = search.data.results or []# 2. Ask the same questions of every video, all at onceschema = { "product": {"type": "String", "description": "Product being discussed"}, "main_claim": {"type": "String", "description": "Main claim made about it"}, "sponsored": {"type": "Boolean", "description": "Is it a paid partnership?"},}jobs = [client.submit_extract_video_data(video_url=v.url, schema=schema) for v in videos]# 3. One row per videowith open("creatine-gummies.csv", "w", newline="", encoding="utf-8") as f: writer = csv.writer(f) writer.writerow(["url", "views", *schema]) for video, job in zip(videos, jobs): try: fields = job.result().data except VidNavigatorError: continue # e.g. a video with no speech views = video.stats.views if video.stats else "" writer.writerow([video.url, views, *(fields.get(k) for k in schema)])
Outputcreatine-gummies.csv
url,views,product,main_claim,sponsored
tiktok.com/@…/video/77064…,221100,creatine gummies,"Many creatine gummies contain little or no actual creatine",False
tiktok.com/@…/video/77042…,36100,creatine gummies,"Same creatine monohydrate as powder, no mixing needed",
tiktok.com/@…/video/77047…,16100,creatine gummies,"Six gummies a day deliver 3,000 mg of creatine",False
tiktok.com/@…/video/77062…,2177,creatine gummies,"Helps strength, performance and recovery",False
tiktok.com/@…/video/77057…,146500,supplements,"Prices go up on Monday, buy before then",False
…
VidNavigator is a video intelligence API and MCP server. It searches, transcribes, analyzes and extracts structured data from online videos and uploaded files, so developers, AI agents and data teams can work with video the way they work with text.
Which platforms are supported?
YouTube, TikTok, Instagram, Facebook, X (Twitter), Vimeo, Dailymotion, Loom, Rumble and Snapchat, plus your own audio and video files (mp4, mov, mkv, mp3, wav and more).
Can my AI agent use it?
Yes. The hosted MCP server at https://api.vidnavigator.com/mcp/ works with Claude Desktop, Cursor, n8n and any MCP client that supports HTTP servers. It exposes 17 tools covering video search, transcripts, speech-to-text, analysis, data extraction, TikTok search and your uploaded files.
How do I process many videos at once?
Use the async endpoints: each call returns a task ID immediately, the jobs run in parallel on our servers, and you collect the results by polling or with a signed webhook. The Python and JavaScript SDKs do the submitting, polling and pagination for you.
How long can a video be?
Synchronous endpoints accept media up to 10 minutes. Async jobs have no length limit, so the same code works for a 20-second reel and a three-hour stream.
How does pricing work?
Every plan includes a monthly pool of credits shared by all endpoints. You can start with a free 14-day trial, no credit card required, then choose a plan or contact us for volume pricing.
Do I need to write code?
No. You can use VidNavigator from an n8n workflow, the Chrome extension or the Studio web app. The API and SDKs are there when you want to automate further.
Start with one video. Scale to a million.
Create an account, copy your API key and make your first call in under five minutes.
YouTube
TikTok
Instagram
Facebook
X
Vimeo
Dailymotion
Loom
Rumble
Snapchat
