youtube-agent

YouTube Agent

Platform Agent · run inside Claude, ChatGPT, Claude Code, Cursor, Codex, OpenClaw, Hermes, or any MCP client. Read-only public-data research — eyes, not hands.

The YouTube Agent turns public YouTube into channel and content research: channel and video metadata, view-count signal, related videos, search, and trending. It owns the YouTube Data Skill — the deterministic read path — and composes the Role-Agent Skills behind creator vetting, competitor demo reception, and content-demand work — and never uploads or comments.

Powered by: Influencer marketing · Competitive intel · Content strategy

Who it's for

• Influencer teams vetting YouTube channels before spend
• Competitive teams reading demo and launch reception
• Content teams sizing video demand for a topic

Skills in this agent

• youtube — The deterministic guide to reading public YouTube data through UnifAPI — channels, videos, related content, search, and trending. (data)
• kol-pricing — Estimate Twitter/X KOL rates from public engagement, audience fit, and risk. (enhanced · shared)
• creator-shortlist — Find creators by niche, audience language, recent momentum, and platform fit. (enhanced · shared)
• audience-fit-check — Vet a single creator's audience fit and brand-safety from public posts and engagement before outreach. (enhanced · shared)
• competitor-profiling — Turn a competitor's public footprint into a structured, source-cited profile — positioning, product, pricing, and implications. (enhanced · shared)
• content-opportunity-brief — Find repeated questions and formats across YouTube, TikTok, Reddit, and X. (enhanced · shared)

What you get

• Channel and video metadata with view/like counts
• Related-video and trending signal for a niche
• Search-based demand for a topic or format
• Each figure cited to the public video or channel

Install & run

Connect in a couple of steps — prompt, web connector, or plugin, all read-only with OAuth sign-in (new workspaces get free trial credits):

Paste one prompt (Claude Code, Codex, Cursor, OpenClaw, Goose — any agent that manages its own MCP servers): ask it to add the UnifAPI MCP server at https://mcp.unifapi.com — a remote streamable-HTTP, read-only server — and run the OAuth sign-in. No API key to paste.

Web connector — no terminal (Claude, ChatGPT, Perplexity, Grok): add a custom connector pointed at https://mcp.unifapi.com, then authorize over OAuth.

Add the skills in any client with npx skills add unifapi-agent/agents. Claude-compatible hosts (Claude Code, Claude Desktop, OpenClaw) can instead install skills + MCP together as a plugin: /plugin marketplace add unifapi-agent/agents then /plugin install unifapi@unifapi.

Step-by-step for every client → https://unifapi.com/mcp

Then ask your assistant:

Research YouTube for my topic: pull the channels and videos that own it, their view and like counts, and related content. Return a cited brief on demand and the strongest channels — read-only, do not upload or comment.

Reads from

UnifAPI APIs: youtube — public data only.

This is a read-only research agent for one platform. It reads public data and returns evidence — it never logs into, posts to, comments on, or messages on the platform, and it never holds your account credentials.

FAQ

Can it read YouTube comments?

Not in this catalog — YouTube exposes channel and video metadata, view/like counts, related videos, search, and trending, but no comment listing. The agent uses titles, descriptions, and counts as the demand signal.

Does it upload or comment?

No. The YouTube Agent is read-only research; you publish anything from your own account.

Links

• Agent page: https://unifapi.com/agents/youtube
• Install & all clients: https://unifapi.com/mcp
• MCP server: https://mcp.unifapi.com
• All agents: https://unifapi.com/agents · Docs: https://docs.unifapi.com

_{Topics: YouTube research tool · YouTube channel research · YouTube creator research · YouTube research agent · YouTube competitor research · YouTube MCP server}