MCP server for AI agents

Let your agent make the media
and post it.

Your agent asks for an image or a video. SuperGen renders it, publishes it to 9 platforms and schedules the rest — one hosted server, no SDK, no browser automation to babysit.

25 free credits, no card · bills only on success

Works with any AI agent

Claude CodeClaude DesktopChatGPTCursorOpenClawHermes
MCPREST APIBearer auth
your agent → supergen
generate_video(
  prompt="paper boat in a stream after rain",
  aspect_ratio="9:16",
)
→ { job_id: "job_ab12", status: "queued" }

get_job("job_ab12")            # free
→ { status: "succeeded", assets: [...] }

create_post(
  account_id="acct_x",
  media_urls=[...],
  scheduled_time="2026-08-01T09:00Z",
)
→ { status: "scheduled" }
X (Twitter)
scheduled
YouTube
scheduled
LinkedIn
scheduled
Instagram
scheduled
TikTok
scheduled
Facebook
scheduled
Threads
scheduled
Telegram
scheduled
Bluesky
scheduled
41
MCP tools
9
platforms
11
models
25
free credits

The whole loop.
One server.

Generating the media and getting it published are the same job. Splitting them across four subscriptions is what makes it a chore.

Generate

Image and video, routed for you.

Name a model or let SuperGen pick — cheapest capable, best in class, or a balanced middle. References keep a character or style consistent across runs.

Publish

One call, the right format.

Each platform has its own caps, media rules and required fields. Send the post once; SuperGen shapes it per platform and refuses what would be rejected instead of failing halfway.

Schedule

Your agent does not wait up.

Hand over a time and close the session. Scheduling runs server-side, and a retry never turns into a second post.

Compose

More than one model can make.

Templates fan out several generations and assemble them — slideshow with voiceover and captions, clip reels, quote carousels. Assembly is deterministic and never bills credits.

Measure

What actually landed.

Live likes, comments and views per post, and your top performers grouped by the preset that made them — so weak presets get retired on data.

Inbox

Replies belong in the loop.

Comments and DMs on what you published, readable and answerable by the same agent that posted it.

Connect in one line

Point your agent at a URL.

Streamable HTTP, bearer auth, nothing to install. The same key works for the MCP server and the REST API.

Setup for every client →
claude mcp add --transport http supergen \
  https://mcp.supergen.tech/mcp \
  --header "Authorization: Bearer sk_…"

Real questions.
Straight answers.

01Do I need to install anything?+

No. SuperGen is a hosted MCP server. You point your agent at a URL and give it your API key — no SDK, no browser automation to babysit.

02What exactly costs credits?+

Only generation, and only when it succeeds. A failed render bills nothing, polling a job is free, and publishing, scheduling, analytics and inbox calls are free. Retrying with the same idempotency key returns the original job rather than rendering twice.

03Can I use it without an agent?+

Yes. The same actions are in the web app: compose to several accounts at once, preview per platform, and a calendar of what is queued. The browser is a second door onto the same service, not a second product.

04Which models can I use?+

The live catalogue with the credit cost of every model is on the Models & prices page. You can pin one by id, or state a preference and let the router choose.

05Does it declare AI-generated media?+

Yes, automatically. Where a platform requires it — TikTok, YouTube — posts built from a generated job are flagged as synthetic without you passing anything. Undeclared AI media gets down-ranked, so this is not optional.

Point your agent at SuperGen.

41 tools, 9 platforms, 25 credits to try it. Bills only on success.

Start free →