Adspirer

About Adspirer

Adspirer is an MCP server that puts paid-media work inside the assistant you already use, so campaign analysis and campaign creation happen in a Claude, ChatGPT, Cursor, Codex or Windsurf conversation rather than in six ad platform dashboards. It connects Google Ads, Meta Ads, LinkedIn Ads, TikTok Ads, Amazon Ads and ChatGPT Ads, and the pricing is deliberately decoupled from ad spend: you pay per task the agents perform, never a percentage of budget, so a performance pull is one task and a full multi-platform launch is several. The half that makes it interesting to developers rather than only to marketers is the open plugin ecosystem. Adspirer plugins are Apache-2.0 MCP tools written in Python: you decorate a function with @tool, declare the read scopes it needs, and reach account data through ctx.broker — never writing auth code, never seeing an OAuth token, never touching a database. There are 163 read-only tools to compose against for clustering, scoring, anomaly detection, recommendation and reporting. The security architecture treats plugin code as untrusted, which is the right default: plugins run in a network-isolated process with no credentials, can only read through the broker, cannot modify campaigns at all in v1, and the broker enforces server-side that a plugin only ever sees the invoking user's own data. Every pull request passes static analysis, secret, dependency and malware scans, schema-quality checks and human maintainer review before merging, after which the tool is live for all users.

Key Features

Tags