Social Magnum can be connected to third-party AI agents and assistants (for example Claude, ChatGPT, Cursor, Codex or any other client that speaks the Model Context Protocol) through our MCP connector. Connecting one is entirely optional, and no agent can reach your account until you approve it yourself on our consent screen while signed in.
An approved agent acts as you, and you are responsible for everything it does. An agent you connect can publish posts, reply publicly to comments and messages, schedule and cancel posts, upload and delete media, create and delete drafts and auto-reply rules, and switch between your workspaces. Anything it does is treated as your own action, exactly as if you had performed it in the App yourself.
- Publishing is immediate and cannot be undone. When an agent publishes a post or a reply, it goes out to the connected social network straight away. Deleting it afterwards removes our copy and asks the network to remove theirs; it does not recall what people already saw, and it does not undo the post having existed.
- We do not review, approve or filter what an agent creates or sends. We are not the author of that content and we do not check it before publication. Section 7 (Content Responsibility) applies to agent-generated content in full.
- The agent is not our product. Third-party AI agents are operated by their own providers under their own terms and privacy policies. We do not control their behaviour, their accuracy, their availability, or how they handle the data they read from your account. We make no warranty about any agent's output.
- AI output can be wrong. Agents can misread instructions, invent details, pick the wrong account, or act on text they read inside a comment or a message. You are responsible for reviewing what you ask an agent to do and for the consequences of granting it access.
- Platform rules still apply. Posts published by an agent are subject to the connected network's terms in the same way as any other post, including any restrictions that network places on automated or AI-generated content. Section 5 applies unchanged.
To the fullest extent permitted by law, we are not liable for any loss, damage, cost or claim arising from the actions of an AI agent you connected: including posts published, replies sent, content deleted, or data read. Whether or not those actions matched what you intended. This does not limit our liability where the loss is caused by our own failure to operate the connector as described in these terms.
You can see every connected agent, and disconnect any or all of them, from the Agents & CLI page in the App. Disconnecting takes effect immediately: the revoked credential stops working on its next request. It does not undo anything the agent already did. You are responsible for keeping this list current and for removing agents you no longer use.