For AI agents
An AI marketer needs safe hands.
One MCP server or REST API gives your agent your social accounts and your marketing data, with guardrails built in: it checks every post, asks before posting and never posts twice.
tools/list
- list_accounts
- get_platform_rules
- upload_media
- create_post
- get_post
- list_posts
- cancel_post
- list_account_tools
- run_account_tool
Guardrails
Autonomy with limits.
Built into the workflow every post goes through, whether it comes from an agent, your code or the dashboard.
Checked before it’s sent
Never posts twice
One key per agent
Scheduling and cancel
Data, not just posts
An AI CMO loop
From data to post.
Read what’s working, draft, confirm, publish, check. Every step is a tool call.
-
1
run_account_tool
Pull a report from Google Search Console.
-
2
get_platform_rules
Check how long each platform lets a post be.
-
3
create_post
Publish the approved draft to the chosen accounts.
-
4
get_post
Confirm it’s live and collect the links.
FAQ
Agent questions.
MCP or REST for my agent?
If your agent runs in an MCP client (Claude Code, Cursor, VS Code and others), use the MCP server: it’s tools with descriptions, no code. If you’re writing your own agent, call the REST API or connect it to the MCP endpoint with an MCP client library.
Can an agent post without me?
It can, if you let it: the API key allows posting. The server asks agents to confirm with you first, and you choose the accounts the workspace connects. For unattended agents, use a dedicated workspace with only the accounts it should use.
How fast can an agent go?
120 requests a minute per API key, shared by REST and MCP. Over the limit returns 429 with a Retry-After header.
Can the agent read results back?
Yes. get_post returns each account’s status and live URL, and data tools return reports from Facebook, Instagram, Google Search Console and more.
Give your agent the keys.
Create a free account, connect an account and send your first post in a few minutes.