Blend MCP

looks complicated, bright UI may better

Hey Fazier 👋 We built Blend MCP because we were tired of jumping between 8 ad dashboards. Now you just ask Claude or ChatGPT how your ads are doing, where you're wasting spend, or to tweak a campaign - and it does it. Would love your feedback!

We run a hosted MCP server for publishing social posts, so I’m curious how you handle write access to ad accounts, where a wrong call costs real money. Do you split read and write permissions on the OAuth screen, and do budget or status changes require a confirmation argument before they execute?

The single remote MCP server with OAuth sign-in and the official MCP Registry listing is a sensible setup for ad accounts. One suggestion: a read-only scope or a per-platform permission toggle on the OAuth consent screen would let an agent pull reports without campaign-edit rights, which is probably the first question agencies will ask.

Having one remote MCP server cover eight ad platforms is a clear setup. I'm curious how it behaves when a platform's API hits a rate limit or quota error in the middle of a long report: does the assistant get a plain error it can explain, or does the server retry quietly? Documenting that would help people judge how reliable multi-platform reports are.

Eight ad platforms behind one MCP server with OAuth is the part I like, LinkedIn and Amazon are usually the ones left out. For agencies juggling lots of accounts, which AI client do most of your paying users actually run it from, Claude or Cursor?

Covering Reddit and Pinterest alongside Google and Meta in one MCP server is handy for cross-platform reports. Is there a dry-run or confirmation step before budget or bid changes are applied, so a misread prompt doesn't change live spend?

Having Google, Meta, TikTok and LinkedIn behind one MCP server is genuinely useful. For budget changes I'd love a required "preview, then confirm" step so an assistant can never move spend on its own. Is that configurable per account?

Write access is the part I'd want to see spelled out. Does it ask for confirmation before budget or bid changes, and is there a log of what the assistant changed per account? For agencies that audit trail matters more than the reporting.