Agents and MCP

Social publishing over MCP, governed by humans.

RostrumPost exposes a REST API, a CLI, an MCP server, and an agent Skill that share one authorization model. Give an agent a draft-only key and require approval before anything publishes.

Start free Read the API

SurfacesAPI, CLI, MCP, Skill

KeysScoped

Draft-onlyYes

ApprovalPer workspace

AuditExportable

What you get for agents

Scoped keys

Grant only the scopes a workflow needs, from read-only to publish.

Draft-only mode

Agents compose and preflight without the ability to publish.

Approval required

Gate publishing behind human approval per workspace.

Audit log

Every key, run, and mutation is recorded and exportable.

How agent publishing stays safe

An MCP server is only safe if the agent behind it cannot quietly publish. RostrumPost keeps publishing and scheduling behind explicit scopes and human confirmation, and records every action.

The same authorization model covers the REST API, the CLI, and the agent Skill, so a draft-only policy holds no matter how the agent connects.

Draft, preflight, then wait for a human

Publishing and scheduling require explicit scopes plus confirmation, on every surface.

rostrum cli

$ rostrum post --caption "Launch copy" --platforms x,linkedin
> preflight: 2 direct, 0 blocking
> variants ready: x, linkedin

$ rostrum publish preview <variant_id>
> status: ready
> requires scopes: jobs:write + publish:write
> awaiting human confirmation

MCP and agent questions

Is there an MCP server?

Yes. The MCP server shares the same scoped authorization model as the REST API, CLI, and agent Skill.

Can an agent publish without approval?

Only if you grant it. You can keep agents draft-only and require human approval before anything publishes.

What is recorded?

Every key, agent run, and mutation is written to an audit log you can export.

Publish with fewer surprises.

Turn prepared content into native variants, preflight every channel, and publish only where the connection allows.