What happened
In an article dated November 4, 2026, Anthropic announced new partner agents and products for Claude Marketplace. The marketplace is described as a place where teams can find partner tools and services to use with Claude. Anthropic says there are now more than 25 partner agents and products listed. Teams can browse the listings and request to buy from any of them. Eligible enterprise customers can apply a portion of their committed Anthropic spend toward these solutions, so one commitment can cover Claude and partner products, whether a customer is buying something new or expanding use of a platform it already has.
The additions span several categories. In security, Chainguard Harness builds applications from source in an isolated microVM and probes the live application for exploitable vulnerabilities; Anthropic says it can offer custom remediation recommendations for reproducible vulnerabilities. Checkmarx One combines rules-based analysis with AI reasoning to identify code vulnerabilities and help teams prioritize and fix findings before release. Cyera Agent Guardian inventories Claude deployments, governs permissions before Claude runs, and enforces policy on prompts, files and tool calls; it also supports testing those controls through red teaming. Rubrik Code Guardian uses frontier AI models from Anthropic to review a secure copy of GitHub or Azure DevOps repositories for multi-step attack paths, rank issues by severity and turn confirmed critical issues into tracked tickets. Tanium Cloud provides real-time visibility and control across endpoints to help find and address vulnerabilities, misconfigurations and missing patches.
For engineering infrastructure, Datadog Agent Observability traces agent execution through tools and services, with the stated aim of helping teams improve quality, performance and cost efficiency by surfacing production patterns and recommending fixes. Netlify Enterprise deploys apps, sites and agents to a global edge network and supplies preview URLs for each change. Anthropic describes it as a route from Claude-built prototypes to production without infrastructure configuration; the listing also includes SSO, role-based access and audit logs.
The creative and marketing additions are AirOps, Runway and Webflow. AirOps supports AI-search strategy, content and multichannel execution, and can connect brand performance in AI search to pipeline and revenue. Runway generates and edits video, images and audio, including natural-language editing of footage. Webflow provides tools to build, manage and optimize websites, with controls and governance for marketers, designers, developers and agents.
Productivity listings include Asana AI Teammates, prebuilt agents drawing on the Asana Work Graph, each with an identity, scoped permissions and an audit trail; Asana says they route decisions requiring judgment to an owner. Glean connects company knowledge and workflows across systems. monday agents work within projects, boards, documents and meetings, using owners and deadlines as context and offering controls over access, approvals and spending. Zoom Agentic Search answers questions inside Claude using meeting summaries, work conversations, documents and connected third-party sources; Anthropic says its results are permission-aware and do not require analysis of raw transcripts or tool chaining. The article does not give separate prices, rollout dates or technical integration instructions for these products.
What the archive adds
This is a new fact on a known story, not a new marketplace launch. Our earlier Field Note on Claude Marketplace described the marketplace as joining technical extensions, enterprise procurement and implementation services. The present announcement adds named products across categories and specifies that eligible enterprise customers can use a portion of committed Anthropic spend for partner solutions. The source does not say whether the marketplace's overall commercial terms or access have changed.
Trend signal
The update offers a modest signal for agent-ready software exposes capabilities: several listed offerings emphasize scoped permissions, policy controls, audit trails or permission-aware access. That fits the pattern's focus on observable, governed capabilities that agents can use. It does not establish that these controls are standardized across the marketplace, nor does it show how they are implemented or evaluated in practice.
Why now
Anthropic's article frames the additions as a way for teams to get more from Claude and emphasizes applying committed spend across Claude and partner products. It does not explain why the announcement is timed for November 4 or identify a market event that triggered it, so no broader timing claim is warranted.
Connections to other events
A prior Telegram post in the archive described Claude Marketplace as a unified catalog for connectors, plugins, agents, products and service partners, with more than 2,000 integrations reported at launch. The current announcement is narrower: it names newly added products and says there are now more than 25 partner agents and products. These figures describe different things—integrations versus partner agents and products—and the source does not reconcile them.
Why it matters
For enterprise teams, the practical question is shifting from whether a marketplace exists to whether a listed partner product fits a real workflow and can be governed within existing controls. The examples span concrete needs: reviewing repositories for attack paths, tracing agent runs, deploying applications, and searching work content with permissions in view. That breadth may make the marketplace useful as a discovery and procurement surface, particularly where teams already have an Anthropic commitment.
But a catalog entry is not evidence of a successful deployment. Buyers still need to check which systems a product connects to, what data it can access, who approves its actions, how findings are verified, and how costs are charged. The announcement supplies short descriptions, not comparative evaluations, implementation details or customer results. Teams should treat the listings as options to assess, not as proof that an agent is safe, effective or cheaper than their current workflow.
Limits and uncertainty
The source is Anthropic's own marketplace announcement, and its descriptions are not independent evaluations. It does not provide prices, specific availability dates, product-by-product eligibility rules, customer adoption figures or measured outcomes. It also does not explain how much committed spend may be applied, how purchases are invoiced, or whether all listed products are available in every region or enterprise plan. Product descriptions identify intended functions but do not establish their performance, security properties or interoperability in a customer's environment. The “over 25” figure refers to partner agents and products; it should not be conflated with the separate earlier marketplace claim about integrations.
Skill to build now
Before trialing one of these listings, write a short evaluation card for a single workflow: the system it must connect to, the data and actions it requires, the human approval point, and the evidence that would count as success. For a security product, that might mean testing a known repository issue and checking whether its finding can be reproduced and tracked. For an observability tool, define which agent runs and cost signals you need to see. This makes marketplace discovery answer a team-specific question rather than becoming a tour of vendor claims.
Reader takeaway
Claude Marketplace is growing as a partner-product and procurement layer, with new options spanning security, engineering, creative work and productivity. The announcement makes the catalog more concrete, but does not prove product quality or spell out the commercial details. Teams should evaluate a listing against a bounded workflow, its permissions and measurable outcomes before treating it as an operational capability.