--- title: "Cork vs Cursor Projects: a coordinator for Cursor agents, or shared state for every agent · Cork" description: "Cork vs Cursor Projects: a Projects coordinator plans work and directs Cursor's agents with shared context files; Cork keeps any agents on one codebase aware of each other." canonical: "https://corkcli.com/blog/cork-vs-cursor-projects" last_updated: "2026-11-14" published: "2026-10-19" author: "Kevin Vera" --- # Cork vs Cursor Projects: a coordinator for Cursor agents, or shared state for every agent Cork vs Cursor Projects: a Projects coordinator plans work and directs Cursor's agents with shared context files; Cork keeps any agents on one codebase aware of each other. By Kevin Vera · October 19, 2026 · [Comparisons](https://corkcli.com/blog/c/comparisons) · 9 min read How is Cork different from Cursor Projects? Cursor Projects, launched on September 10, 2026, gives a larger body of work (a feature, a migration, a full app) to a coordinator agent that plans it, delegates the pieces to Cursor's own cloud and local agents, and keeps a set of shared context files in sync across every machine those agents use. Cork does not host or direct agents: it sits under the agents you already run (Claude Code, Codex and Cursor) and keeps them, and the people behind them, from overwriting each other on one codebase. The short version: Projects decides who does what inside Cursor; Cork makes every agent on the repo aware of what the others are changing while they write. We already compared Cork with [Cursor cloud agents](https://corkcli.com/blog/cork-vs-cursor-cloud-agents). Projects is a separate, newer layer on top of them, so it gets its own page. ## Cork vs Cursor Projects at a glance | Topic | Cork | Cursor Projects (as of October 2026) | | --- | --- | --- | | What it is | [Coordination layer](https://corkcli.com/blog/coordination-layer-ai-coding-agents) for multiple AI coding agents and teammates on one codebase | A coordinator agent that plans a larger body of work and directs other agents to do it | | Who plans the work | You, or whatever runs your agents | The Project's coordinator agent, which "doesn't write code itself" | | Agents covered | Claude Code, Codex and Cursor via hooks, plus any MCP client | Cursor's own cloud and local agents, as Cursor's pages describe it | | Where it runs | Your agents run where they already run; hooks on your machine | "A Project runs on its own computer in the cloud" | | What is shared | Signatures, callers, recent changes, notes and messages between live agents, per symbol | A set of Project files that sync across every cloud and local machine its agents use | | Triggers | Normal agent prompts; hooks brief the agent automatically | Chat with the coordinator; it can watch a Slack channel, run on a schedule, or follow your PRs | | Conflict handling | Write gate, live awareness, [semantic lint](https://corkcli.com/blog/semantic-lint), rebase-and-test landing | Planning and delegation by the coordinator; cloud agents' work comes back as branches and PRs | | Pricing | Free (local, and 10 agent hours a month in the cloud); $20 a month per 100 agent hours, pooled | On every plan that includes Cloud Agents (Premium and higher, Teams, Enterprise); no separate Projects price listed | ## What is Cursor Projects? Cursor announced Projects on September 10, 2026: "Projects lets you take on larger bodies of work, such as a feature, a migration, or a full app." You run a Project by chatting with its coordinator agent: "The coordinator doesn't write code itself but directs other agents that do." Cursor's changelog puts it this way: "it plans the work, delegates it to agents that implement it, and brings the finished work back to you." Two things make it more than a single long-running agent: - **Shared context.** "Each Project maintains a set of files that sync across every cloud and local machine its agents use." Every agent working on the Project reads the same notes. - **Triggers.** "The coordinator can watch a Slack channel, run on a schedule, or follow all your PRs." A Project lives in Cursor's cloud ("A Project runs on its own computer in the cloud, so closing your laptop doesn't stop it"), and Projects launched in beta, "rolling out to all users". Cursor's docs say it "is available on every plan that includes Cloud Agents: Premium and higher, Teams, Enterprise, and legacy paid Cursor plans." Cursor's docs say the coordinator "creates and manages agents on your behalf, running as many in parallel as the work needs," and "when something needs testing on your machine, the coordinator starts a local agent to run it there." Cursor's announcement, changelog and docs describe it delegating to Cursor's cloud agents and local agents. None of them mentions Claude Code, Codex or other outside agents as workers. (Cursor's own agents can run on other companies' models; this is about which agent harness does the work, not which model.) ## What is Cork? Cork (corkcli.com, by Lithometric) keeps several AI coding agents, and the people running them, from overwriting each other on one codebase, and hands each agent what the others already learned. It does not host agents or machines: your agents run where they already run. Hooks in Claude Code, Codex and Cursor do the protocol with no extra model calls, and an MCP server (https://mcp.corkcli.com) serves any MCP client. The hooks send the structure of your code (symbol names, signatures, docstrings, line hashes), never the file text. - **Briefing**: the exact signatures, callers and recent changes of the code an agent is about to touch, delivered at start and with each prompt. - **Live awareness**: a rename or new signature from another agent reaches yours on its next step. - **[Write gate](https://corkcli.com/blog/write-gate)**: an edit against code another agent is changing right now is held until your agent has their version. - **Semantic lint**: teammates' renames and signature changes become lint rules. - **Landing**: `git push` rebases onto teammates' commits, reruns tests and pushes in one step. - **Agent messaging** anchored to code, a live code graph and a team dashboard (cloud plans), and MCP tools like get_symbol, blast_radius, recap and remember. ## A plan is not a shared state Projects and Cork both answer "how do many agents work on one thing", from opposite ends. A coordinator works top-down. It splits the work before anyone starts, hands each agent its piece, and the shared Project files carry the plan and what has been learned. That works well when the pieces really are separate. It says nothing about the moment two agents' pieces turn out to touch the same function: the plan was made before that function changed. Cork works bottom-up, at the level of the code. It does not know your plan; it knows that agent A renamed `parse_config` three minutes ago, that agent B's next edit calls the old name, and that agent C is mid-change on the same function right now. It tells B before B writes, and holds C's edit until C has A's version. It also covers agents nobody planned together: a teammate's Claude Code session, a Codex run in another terminal. ## Where Cursor Projects stands out - **Planning and delegation built in.** You describe a feature or migration once; the coordinator breaks it down and runs it. Cork does neither. - **Runs without your laptop.** A Project keeps going in Cursor's cloud. - **Triggers.** A Slack channel, a schedule or your pull requests can start or steer work. - **One place to steer.** You talk to one coordinator instead of many agents. ## Where Cork wins - **Any agent, not one vendor's.** Cork coordinates Claude Code, Codex and Cursor in the same repo, plus any MCP client. Cursor's pages describe Projects directing Cursor's own agents. - **Symbol-level, live state.** Shared files carry what agents chose to write down. Cork's briefing and deltas come from the code itself: who changed which symbol, its new signature, its callers, as it happens. - **Collisions held before they land.** The write gate stops an edit against code another agent is changing right now; Projects relies on the plan having kept agents apart. - **People and agents together.** Teammates on their own machines, with their own agents, see the same state (paid plan). - **Measured.** In Cork's [Study 5](https://corkcli.com/benchmarks/study-5-twelve-agents-53-percent) (12 Claude Code agents at once on one repository, hard tasks), agents with Cork used 53% fewer tokens than the same agents without it. ## Can you use both? Partly. Cork's hooks work with Cursor on your machine, so your local Cursor agents, alongside any Claude Code or Codex agents, are briefed and gated against each other. A Project's cloud agents run on Cursor's machines, and we do not claim Cork runs inside those. A workable split: give a Project well-separated work and let its pull requests come back; keep the busy, shared code (interfaces, renames, cross-cutting refactors) with local agents under Cork, whose landing step rebases onto the Project's merged work and reruns your tests before pushing. ## Pricing compared Cursor's docs say Projects is available on every plan that includes Cloud Agents (Premium and higher, Teams, Enterprise, and legacy paid plans). We did not find a separate price for Projects on Cursor's pages. Cork Free runs unlimited agents and repos on one machine with no account, with all data on the machine. Cloud Free gives each account 10 agent hours a month (30 bonus hours the first month), unlimited repos and agents, and a 30-day history window. Beyond that you pay for agent hours, not seats: $20 a month per unit of 100 hours, pooled across the workspace, graduated to $17 (units 11 to 50) and $15 (51 and up), with 15% off a year. Every member is a player, with teammates' machines, prompt-to-code matching by meaning, shared knowledge and 90-day history. Enterprise starts from about 100 units. ## Bottom line Cursor Projects is a strong way to hand Cursor a large, plannable piece of work and let a coordinator run it in the cloud. It coordinates by planning. Cork coordinates by sharing live state: whatever agents you run, wherever you run them, each one knows what the others are changing in the code it is about to touch. If you want a manager for Cursor agents, try Projects. If your agents, from any vendor, keep stepping on each other, add Cork. ## Frequently asked questions ### Does Cursor Projects work with Claude Code or Codex? Cursor's announcement, changelog and docs describe the coordinator directing Cursor's own cloud and local agents, and do not mention Claude Code or Codex as workers. Cork coordinates all three. ### What are Cursor Projects' shared context files? Each Project keeps a set of files that sync across every cloud and local machine its agents use, so all of them read the same notes. ### Is Cork a coordinator agent? No. Cork does not plan or assign work. It gives each agent the live state of the code it is about to touch and holds edits that would collide. ### Can Cork and Cursor Projects run on the same repo? Yes. Local agents under Cork and a Project's cloud agents can work on one repo; cloud agents' work arrives as branches and pull requests, and Cork's landing step rebases local work onto it and reruns tests. ### When did Cursor launch Projects? September 10, 2026, in beta. ## Sources Checked October 19, 2026: - (September 10, 2026) - - - - - - ## Related reading - [Cork vs Multica: assigning agent work vs keeping it from colliding](https://corkcli.com/blog/cork-vs-multica): Cork vs Multica: Multica assigns issues to AI coding agents on a shared board. Cork keeps parallel agents from overwriting each other's code. - [Cork vs Amp: one agent in many orbs, or many agents on one codebase](https://corkcli.com/blog/cork-vs-amp): Cork vs Amp: Amp runs its agent in cloud orbs you can share. Cork keeps Claude Code, Codex and Cursor agents from overwriting each other on one repo. - [Cork vs Factory: a coordination layer, not another agent platform](https://corkcli.com/blog/cork-vs-factory): Cork vs Factory: Factory's Droids plan, build and ship code across your SDLC. Cork keeps parallel AI coding agents from overwriting each other's work. - [Cork vs Amika: VMs for your agents, or coordination between them](https://corkcli.com/blog/cork-vs-amika): Cork vs Amika: Amika provisions multiplayer VM Rigs for coding agents. Cork keeps parallel agents on one codebase from overwriting each other. [All articles](https://corkcli.com/blog.md) · [Cork in Markdown](https://corkcli.com/index.md) · [llms.txt](https://corkcli.com/llms.txt) This page in HTML: https://corkcli.com/blog/cork-vs-cursor-projects