Cursor Team Usage Analytics: Dashboard, API & AI Code Tracking

Team admins see data for themselves and every member. Members without admin rights see their own data and, in some views such as the Usage Leaderboard, data for selected teammates. Every member can open the leaderboard, which always shows the top ten users. Analytics only collect data from clients on version 1.5 or later.

Each chart has a CSV download, and an icon in the page header downloads every chart at once.

API access is where Cursor's pages disagree. Its analytics docs say programmatic access through the Analytics API is for Enterprise customers only, while its team pricing page lists usage stats through the Admin API among the Teams features. Confirm which endpoints a Teams admin key can call before you plan an automated report.

Settle who can see whom before you show anyone the dashboard. Peer visibility makes the leaderboard a social object as much as a measurement, so explain what it is for while people are still curious, before somebody finds their own row and draws their own conclusions. Enterprise admins can also restrict the analytics dashboard to admins only.

The failure mode is using any of it in a performance conversation. Acceptance rate measures a habit and a codebase: an engineer working in generated protobufs will score worse than someone building forms, with no difference in skill. Say plainly that the dashboard is not review input, and mean it, because a team that suspects otherwise starts optimizing the chart.

Cursor logs a signature for every AI line, from Tab or Agent, that it suggests during a chat session. It compares those signatures with the lines in later git commits by the same author, and it attributes matching additions and deletions to AI. Detection runs on the device, and Cursor stores only line counts, as metadata for the dashboard and the API.

AI code tracking covers commits from the Cursor CLICursor's command line: the full agent, all modes and models, in the terminal and pipeable into scripts and CI. Press Enter for the full definition. and Cloud AgentsAgents that run in a Cursor-managed virtual machine, check out the repo, do the work and open a pull request, then shut down, with no load on your laptop. Press Enter for the full definition., and the API labels each one with its source. Older guides list both as unsupported. If metrics are missing for a specific workflow, Cursor asks you to report it as a bug.

Read the share as a direction rather than an audit. Attribution starts from what Cursor suggested, so the chart reports code written with Cursor where Cursor could see it. That is close to the share the AI wrote, but not the same thing.

Apply the production-branch filter before you quote the figure outside the team. The filter uses the repo's default branch, or a common default branch name when none is set. The unfiltered view includes throwaway branches, and a skeptical reader can take that number apart with one question.

The dashboard splits usage into charts, and each answers a different adoption question: how much committed code the AI wrote, where edits come from, who the heavy users are and which repos see the most AI activity.

Some chart labels use older names. ⌘+K used to open inline edit and now opens Agent with the selected lines, and Background Agent is the earlier name for Cloud AgentsAgents that run in a Cursor-managed virtual machine, check out the repo, do the work and open a pull request, then shut down, with no load on your laptop. Press Enter for the full definition..

Most of these charts need a monthly read at most, and a couple reward a weekly one. Suggested against accepted, on Agent edits and Tab completions, moves first when something is wrong. If suggestions hold steady while the accept rate falls, check prompting and context before you blame the model. Active users per product answers a blunter question: did a surface land at all?

Messages sent is the easiest chart to over-read, because a chatty week and a productive week look the same on it.

The table leaves out the Cloud Agent charts and Client Versions. The Cloud Agent charts count agents created, pull requests opened and merged, and lines written and merged, with top repositories and top users. Client Versions shows the Cursor version each person last opened that day. Check it before you pin a version in MDM, so you know which builds the team is running.

Conversation InsightsA Cursor analytics view that passively categorises what agents are doing (new features, bug fixes, refactors) so leaders can see where engineering time goes. Press Enter for the full definition. is on by default for Enterprise teams, and admins can turn it off in team settings. It classifies agent sessions by category (bug fixing, new features, refactoring, DevOps and more), by work type, by complexity and by how specific the prompts are. Leaders see what kind of work agents handle without ticket surveys, and Enterprise teams can extend the default categories or define their own.

Compare puts usage side by side across teams and individual developers, so you can spot adoption patterns and find power users.

Raise Conversation InsightsA Cursor analytics view that passively categorises what agents are doing (new features, bug fixes, refactors) so leaders can see where engineering time goes. Press Enter for the full definition. yourself before someone else discovers it. It is on by default, and the classification runs on-device, so the accurate line for a privacy conversation is that sessions are classified locally and the dashboard receives categories.

Check the price before you budget for it. Cursor's docs say Conversation InsightsA Cursor analytics view that passively categorises what agents are doing (new features, bug fixes, refactors) so leaders can see where engineering time goes. Press Enter for the full definition. was free during its preview and that customers are charged for inference from January 1, 2026, a date that has already passed. The Cursor Token Rate also applies when the underlying request goes to an eligible third-party model.

Use the Analytics API for team adoption and product-usage trends. Use the AI Code Tracking API when the question is about commits or individual accepted AI changes. Cursor's docs mark both as Enterprise-only, and both use Basic authentication with the API key as the username. They answer different questions, so keep them separate in your reporting model.

Start in the dashboard even if the data ends up in a warehouse. It lets an admin check quickly that the date window and team population match the question, before an API job turns a mistaken filter into a monthly chart. Once the read is stable, store the endpoint name and filter values with every extract, so somebody can reproduce a number later.

The AI Code Tracking API needs the stronger disclaimer. Cursor labels it alpha, calculates metrics only for the top-level git repository in the workspace root, and does not support multi-root workspaces. A precise commit table built on incomplete workspace coverage can look more authoritative than it is. Keep the limitation visible, and treat field changes as an expected integration event instead of silently accepting a new response shape.

The admin analytics go well past the leaderboard. Read together, they show who uses Cursor, how they use it, and on which models and surfaces. Use them to find power users whose habits are worth copying and stalls worth fixing.

The catalog: what an admin can measure

The reads below are ones you assemble from the dashboard charts and the CSV or Analytics API exports, not a separate set of first-party metrics. They group the same usage data by the questions worth asking after rollout, so you can move from a raw number to a decision. Check current chart names against the analytics docs before you wire any of this into a report.

The read that matters: request volume against acceptance rate

Plot request volume (how much someone leans on Cursor) against acceptance rate (how often suggestions land), and the team sorts into four quadrants. The quadrant shows who to learn from and who to coach: top-right are power users, and bottom-right are people fighting the tool.