NEWS
Cursor's new leaderboard shows teams the most popular plugins, skills and MCPs
A new control plane for agent tooling — with a leaderboard showing what your team is actually using.

Paul Sawers

As engineering teams adopt more agent tooling, keeping track of what's actually running across an organisation has become its own problem. Plugins, skills, and MCP servers get configured differently by different developers, with no shared view of what teammates are using, what's proven out, or what's worth standardising on. The result is a sprawl of JSON config files and scattered settings that nobody has full visibility into.
Cursor's latest update takes aim at that. Version 3.9, released June 22, introduces what the company calls a "Customize" page — a single interface for managing plugins, skills, MCP servers, subagents, rules, commands, and hooks across an organisation, controllable at user, team, or workspace level.
A leaderboard that shows what teammates actually use
The headline feature is a leaderboard showing which plugins, skills, and MCPs are most used both within a team and across the broader Cursor community. For skills, the leaderboard surfaces how many times each has been used by the team in the past 30 days, and what proportion of those invocations were agent-initiated versus human-initiated — useful signal for understanding which skills are genuinely being put to work.
For plugins, teams can see how many teammates have already added a given plugin, and click through to add it to their own setup in one step.
Previously, there was no way to see what teammates had configured — adoption was an individual, manual process with no shared signal. The leaderboard turns it into a discovery surface driven by real usage data, drawing on both internal team behaviour and community-wide trends.
Canvases, shared dashboards, and broader marketplace support
The update also introduces prebuilt plugin canvases — shared, interactive dashboards that render live data from partner tools directly inside Cursor. The Atlassian canvas, for instance, pulls a real-time view of Jira issues, sprint progress, and project documents into the editor, giving teams a live window into their project state without switching context. Teams get a ready-made starting point they can open and reuse, rather than building the wiring themselves.
Team marketplaces, which allow organisations to distribute private plugins internally, now also support GitLab, Bitbucket, and Azure DevOps repositories — previously the feature was limited to GitHub.
Cursor's bigger picture: SpaceX, a GitHub challenger, and a quiet acqui-hire
The update lands at a moment when Cursor is the most closely watched company in developer tools. SpaceX recently confirmed a $60 billion all-stock deal to acquire Cursor's parent company Anysphere — the largest acquisition of a venture-backed startup on record. Around the same time, Cursor unveiled Origin: an agent-native code hosting platform designed as a challenger to GitHub, which has been logging hundreds of incidents over the past year as it struggles to keep pace with the volume of code AI agents are generating.
Elsewhere, Cursor also quietly absorbed open-source coding assistant Continue, in an acqui-hire that shut down the product and handed its codebase to the community under its existing Apache 2.0 licence.
For engineering leaders already managing Cursor deployments at scale, the governance question is only going to grow as agent tooling becomes more embedded in how teams work. A unified control plane and a usage leaderboard won't resolve every challenge, but they give platform teams something they didn't have before: a clear view of what's actually running.
COPY & SHARE

Paul Sawers
Freelance tech writer at Tessl, former TechCrunch senior writer covering startups and open source
READING
·
0%
IN THIS POST
COPY & SHARE

Paul Sawers
Freelance tech writer at Tessl, former TechCrunch senior writer covering startups and open source
YOUR NEXT READ
Why Warp is betting engineering leaders are done picking a favourite coding agent
CEO Zach Lloyd talks multi-harness orchestration, why the CFO is now in the room for AI tool discussions, and what governed autonomy actually looks like on a “factory floor.”

Paul Sawers



