Back to all posts

SocialKit Agent Skills Are Live

Jonathan Geiger
product updateagent skillsai agentssocial mediaapi

We shipped SocialKit Agent Skills.

If your coding agent supports skills, you can now hand it the full SocialKit workflow with one command:

npx skills add SocialKit-API/skills --all

That gives the agent a clean way to work with social media data. It can transcribe a video, summarize it, pull a creator's recent posts, search a hashtag, read the comments, and download a clip, all without you explaining the API in every prompt.

The point is simple: your agent should not need to hand-roll YouTube, TikTok, Instagram, Facebook, X, and LinkedIn scraping logic every time it needs to read a piece of content.

What are agent skills?

Agent skills are small instruction packs for coding agents.

Instead of pasting the endpoints, auth rules, parameters, and response shapes into every prompt, you install a skill once. The agent reads it when the task calls for it and follows the right workflow.

For SocialKit, that means the agent learns how to:

  • authenticate with your SocialKit access key
  • extract transcripts from any supported video
  • generate AI summaries with topics, key points, and quotes
  • research a channel or profile and list its recent content
  • discover videos by keyword or hashtag
  • pull per-video stats and comments for engagement analysis
  • download a video to a hosted file

You stay in control. The skill gives the agent the workflow; your access key and prompt decide what it actually does.

No MCP required

This is the part worth calling out. The skills talk to the SocialKit REST API directly. All the agent needs is a SOCIALKIT_API_KEY. There is no MCP server to run, no local process to babysit, and no extra install beyond the skill files.

If you already run the SocialKit MCP, the skills will happily use those tools too. But MCP is optional. The skills work on their own.

What's in the pack

Installing with --all gives you seven skills:

  • socialkit-api — the shared routing layer: auth, the full endpoint index, and the platform coverage map
  • video-transcripts — transcripts across every platform, single or in bulk
  • video-summaries — AI summaries with topics, key points, tone, and quotes
  • channel-research — profile stats plus a creator's recent videos, reels, posts, or tweets
  • content-discovery — keyword and hashtag search across YouTube, TikTok, and Instagram
  • engagement-analysis — per-video stats and comments, with sentiment you can summarize
  • video-download — download YouTube, TikTok, and Instagram videos to a hosted file

How to use them

Install the skills in your agent environment:

npx skills add SocialKit-API/skills --all

Then add your access key wherever your agent reads environment variables:

SOCIALKIT_API_KEY=your_access_key_here

Grab a key from the SocialKit dashboard under the Access Keys tab. The same key works for REST and MCP.

After that, ask for a concrete outcome:

"Research the @MrBeast YouTube channel: subscriber count and his last 10 videos with view counts, then tell me which topics show up most."

The agent uses the skill to pick the right endpoints, pull the channel stats and recent videos, and summarize the pattern back to you.

A few things I would use it for

  • Turn a competitor's latest TikTok into a transcript and a three-bullet brief.
  • Pull the top comments on a launch video and summarize the sentiment before you respond.
  • Find the top reels for a hashtag this week and rank them by views.
  • Batch-summarize a playlist of YouTube videos and surface the common theme.
  • Download a Reel, transcribe it, and draft a blog post from the transcript.

Each of those is one prompt once the skill is installed.

Skills or MCP?

Use Agent Skills when your coding agent works well with skill files and repeatable workflows, like Claude Code, Cursor, or Codex.

Use SocialKit MCP when you want an MCP-compatible client, like Claude Desktop or another Streamable HTTP client, to call SocialKit tools directly.

Both paths use the same SocialKit API underneath. Skills fit coding-agent environments. MCP fits tool-calling clients. You can run both with one access key.

Try it

Install the skills, add your access key, and ask your agent to pull something real:

"Get the transcript of this TikTok and pull out the three main claims it makes."

Setup notes live in the Agent Skills docs, the skills are on GitHub, and the broader agent setup is on the SocialKit for AI agents page.