Publish your first campaign with MCP
The complete Postward golden path for developers using an MCP-capable coding agent.
The operating model
Your agent creates with your own model and project context. Postward stores the canonical source, versions the outputs, collects approval and publishes.
1. Connect an MCP client
Create your Postward account first. Your workspace is provisioned automatically, then the client opens a browser for OAuth consent on its first authenticated connection.
2. Verify before creating
Check my Postward activation status and tell me the next incomplete step.The agent calls get_activation_status, which verifies the connection and returns an executable next step. Authorization or token refresh alone does not count as a working integration.
3. Submit a reviewable content package
Use the Postward MCP to create a campaign and submit its first content package for review.
Source of truth: inspect this repository's changelog, release notes and recent changes.
Objective: explain what shipped and why it matters to existing users.
Audience: developers evaluating or already using the product.
Destinations: LinkedIn and Discourse.
Before writing, ask only for decisions or facts that are genuinely missing.
Create canonical Markdown first, then generate destination-native materializations from that exact source revision and submit a Run.
Do not invent claims, metrics or customer evidence. Do not approve, schedule or publish.4. Review in Studio
- Open the campaign created by the agent
- Compare each destination version with the canonical source
- Comment or reject when facts, voice or formatting need revision
- Approve the exact Run and destination only when it is ready
5. Connect a destination and publish
Content can be created before a social connection is ready. Connect the destination when you are ready to publish, then publish now or create a schedule from the approved Run.
Next
Build a stronger editorial brief