Claude Marketplace Now Sells Governance Tooling Next to the Agents

Claude Marketplace now sells the controls for agents in the same catalog as the agents. Anthropic’s latest update adds 14 partner agents and products, bringing the total to more than 25. Two of the new listings police other agents rather than act: Cyera Agent Guardian inventories Claude deployments and enforces permissions before Claude runs, and Datadog Agent Observability traces agent execution and recommends fixes for quality, performance and cost.

The security group is the largest. Chainguard Harness builds apps in an isolated microVM and tests them for exploitable flaws, Checkmarx One pairs rules-based analysis with AI reasoning, Rubrik Code Guardian hunts for multi-step attack paths in repositories, and Tanium Cloud covers endpoints. The productivity group — Asana AI Teammates, Glean, monday agents and Zoom Agentic Search — each stresses scoped permissions and audit trails. The article gives no list prices.

The procurement mechanic matters as much as the catalog. Eligible enterprise customers can apply a portion of their committed Anthropic spend to the listings, so one commitment covers Claude and partner tools. That is the model Anthropic set out when it launched the marketplace in September: customers pay for plugins, connectors and Claude-powered products from existing committed spend rather than opening new budget lines.

For a company assembling an agent stack, the practical read is a sequencing question. Observability, permission enforcement and red teaming are no longer a second purchase negotiated after the pilot — they sit on the same order form as the agent. A team that ships an agent without deciding who audits it is now skipping a line item, not a category. Before approving the next agent, name the control that goes with it and the outcome you will measure it against. That pairing is what separates an agent that ships from an agent that stalls in security review.