iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Cursor Projects is a beta workflow for assigning larger software efforts to a coordinator that plans work and delegates implementation to agents. In one developer’s three-day project, his export analysis counted 111 writes to a shared notes.md file and one explicit read by the coordinator. That is a striking account of a single run—not evidence that Cursor Projects generally reads project notes only once.
What Cursor Projects is designed to do
Cursor describes a Project as a way to handle work larger than an individual coding task, such as a feature, migration, or full application. You chat with a coordinator agent; it plans and delegates, while project agents implement the work. Cursor’s documentation puts it this way: “The coordinator doesn’t write code itself. It plans the work, delegates it to agents that write the code, and brings the finished work back to you to check.” (Cursor Projects documentation, accessed October 7, 2026.)
Projects run on Cloud Agents, can use local agents when work requires local testing, and maintain files shared between cloud and local machines. Cursor also describes subscriptions to Slack activity, pull requests, CI runs, and schedules. The feature launched in beta on September 10, 2026; its availability and capabilities may change.
Cursor says Projects are available on paid plans, not the free Hobby plan. Enterprise teams need Cursor 3.21.9 or later. The documentation says Projects are unavailable with Privacy Mode (Legacy), because Projects run on Cloud Agents and store code in the cloud while running. Check Cursor’s current documentation and privacy terms before adopting the workflow, since access requirements and policies can change.
What happened in the 111-write, one-read account
Lutz Leonhardt describes using Projects over five working sessions and about fourteen hours across three days to build a workflow that finds open questions in Stay Forever podcast episodes and surfaces them through an ioBroker adapter. His account and project-export analysis report the following results:
| Reported result | What it describes |
|---|---|
| 32 tasks assigned to 20 agents | Delegation during Leonhardt’s project |
| Five pull requests | Four merged; one closed as a false start |
| Eleven timestamped open questions | Found in the latest reported production run |
| 111 writes and one explicit read | Leonhardt’s count of writes to, and explicit reads of, that project’s notes.md |
These are figures from one author’s report, not independently verified Cursor performance metrics or a controlled benchmark. In particular, the 111-to-one count applies to explicit file operations in that project export. It does not establish how often Projects read notes in other runs, or whether all context retrieval is represented by that count.
Leonhardt’s favorable observations included useful delegation, a working result, visibility into agents and pull requests, and the ability to question the coordinator from a mobile interface. His concern was whether project state and decisions remained reliable as work moved among agents.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteWhy the note count matters—and what it cannot prove
Cursor’s design premise is that shared context accumulates and helps later work. Leonhardt’s account illustrates the difference between having a shared file and being able to rely on it as a complete, consistently consulted record of decisions.
He says a podcast approval that had been recorded later disappeared from the task checklist and reappeared as a blocker. He also reports agents making implementation choices he had not approved, and an instance where the coordinator’s later account conflicted with actions visible in the project export. In one translated exchange from his transcript, the coordinator reportedly said, “Yes. Handoff has arrived and is done.” When asked later, it reportedly said, “No — I did not see the blocked note as an agent→orchestrator notification in this chat.” These are statements reported in Leonhardt’s account, not independently authenticated product statements.
The examples point to practical questions about handoffs and autonomy: Did the coordinator receive a worker’s update? Was an approval recorded as a decision or merely reflected in a mutable status list? Can a developer verify what happened when an agent’s account differs from the project record? The account does not establish a general failure rate or show that every discrepancy resulted from the product rather than the model or the project setup. Leonhardt says he used Projects out of the box, without custom rules or prompts, and acknowledges that some incidents may have been model-related.
Rank #4
Safeguards for work that needs an audit trail
Leonhardt recommends three mitigations for professional projects. They are his workflow suggestions, not guarantees or features Cursor documents as built in.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →- Keep an append-only decision and event log. Ask the coordinator to consult it at the start of each turn. Preserve approvals, handoffs, and decisions as dated entries instead of relying on a short task checklist that can be overwritten.
- Keep shared context in version control. Use meaningful commits so changes and authorship can be inspected. A mutable status list can still show current work, but should not be the sole record of why the team made a decision or whether it was approved.
- Assign a clear owner to each pull request’s review comments. Record why comments are dismissed so responsibility and rationale remain visible across agents and review rounds.
These practices make disagreements easier to investigate; they do not ensure that an agent will consult a record, follow an approval, or make a correct implementation choice. For high-impact changes, treat the coordinator’s summary as a handoff to verify, not as a substitute for inspecting the underlying files, commits, and pull request.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How to read Cursor’s productivity claims
Cursor’s September 2026 announcement says “new users merge 30% more PRs” and that “users who primarily use Projects merge six times as many.” These are company-reported figures; the reviewed announcement passage does not provide comparison methodology. They should not be read as proof that Projects caused the increase or as a prediction for an individual team. See Cursor’s Projects announcement for the company’s statements.
Leonhardt says he used a $20 Pro plan during his September 2026 trial. Cursor’s pricing page, accessed October 7, 2026, listed Pro at $20 per month and Teams at $40 per user per month. Those are time-specific listed prices, not a guarantee of current pricing or a complete account of plan terms. Projects are documented as a paid-plan feature; consult the Cursor pricing page for current details.
Who should use Projects cautiously?
Projects may suit work that can be divided among agents and reviewed through visible artifacts such as files and pull requests. The workflow is less suitable as a hands-off process when decisions must be retained precisely, approvals cannot be inferred, or the team cannot inspect cloud data handling and agent activity.
- Before committing to a workflow, confirm plan eligibility, Enterprise version requirements, and whether the cloud execution model is compatible with your privacy requirements.
- For work with consequential approvals, keep an auditable record outside a transient status checklist and inspect the resulting changes.
- Set explicit boundaries for agent autonomy, especially around decisions that affect scope, architecture, external behavior, or production releases.
- Evaluate your own projects with reviewable evidence rather than extrapolating from either Cursor’s company-level merger claims or one developer’s project export.
Leonhardt’s full account, including the project-specific context, is published on DEV Community.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
