Claude-Ads is an open-source AI skill designed for Anthropic's Claude Code to automate paid media operations. As of August 2026, the tool supports 12 major advertising platforms and focuses on providing source-grounded audits, deterministic scoring, and versioned JSON reports, according to its official repository.
Developed by Daniel Agrici, the project aims to equip marketing agencies, consultants, and in-house performance teams with a reliable AI assistant. It operates by turning authorized account data into actionable audits, plans, and creative workflows. A key feature is its read-only default, ensuring that the AI cannot make live account changes without passing multiple safety gates.
How Does Claude-Ads Automate Marketing?
The system uses a primary conductor agent to manage scope and policy, while specialized worker agents analyze specific parts of an ad account. This modular architecture ensures that if one worker fails, the entire audit is marked as partial rather than presenting incomplete data as a full report. The final output is a versioned JSON file.Users interact with the tool through a series of slash commands. For example, `/ads audit` initiates an evidence-backed review of an account, and `/ads plan` helps build strategies for campaigns, budgets, and measurement. The tool can also draft creative briefs for copy and visuals.
Supported Platforms and Capabilities
Claude-Ads integrates with a dozen of the most prominent digital advertising platforms, covering search, social media, and e-commerce. This wide range of support allows marketing professionals to manage a diverse portfolio of channels using a single, unified AI skill. Supported platforms include leaders like Google, Meta, and Amazon.
Each platform is managed through a dedicated skill, audit worker, and capability manifest. This manifest serves as the official record for what the AI is permitted to read or write, adding a layer of granular control. This structure is similar to other agent-focused projects, such as the open-source AI Research SKILLS from Orchestra-Research, which package capabilities for agents.
Category | Supported Platforms |
|---|---|
Search, Video & Social | Google Ads, Meta Ads, YouTube Ads, LinkedIn Ads, TikTok Ads, Microsoft Advertising, Reddit Ads, Snapchat Ads, X Ads |
Commerce & Retail Media | Apple Ads, Amazon Ads, Pinterest Ads |
Why Is the Safety-First Design Important?
The tool's core design prioritizes account safety by operating in a read-only mode by default. To make any live changes to an ad account, a user must navigate a series of explicit approvals and verification steps. This prevents the AI from making unauthorized or erroneous modifications that could impact ad spend or performance.Applying a change requires an enabled capability for the specific operation, owner approval, and an idempotency key to prevent duplicate actions. Furthermore, the system requires a human-readable before-and-after diff, a rollback plan, and verification that the account state has not changed since the plan was drafted. Credentials are never stored in the repository, profiles, or logs.
