--- title: "Setting up Cork for a team of five · Cork" description: "Set up Cork for a five-person team running Claude Code, Codex and Cursor agents: install, login, agent hours, AGENTS.md, working habits, and what it costs." canonical: "https://corkcli.com/blog/cork-team-of-five-setup" last_updated: "2026-11-14" published: "2026-10-03" author: "Kevin Vera" --- # Setting up Cork for a team of five Set up Cork for a five-person team running Claude Code, Codex and Cursor agents: install, login, agent hours, AGENTS.md, working habits, and what it costs. By Kevin Vera · October 3, 2026 · [Guides](https://corkcli.com/blog/c/guides) · 7 min read To set up Cork for a team of five, have one person buy the team's agent hours with `cork upgrade`, then invite the others from the dashboard, then have each of the five install Cork on their machine with `curl -fsSL https://corkcli.com/install.sh | sh` and link it with `cork login`, choosing the team's workspace. Add Cork's section to each shared repo's `AGENTS.md` with `cork agents-md`, agree on a few habits for messages and notes, and keep your existing branch, review and CI rules. You pay for agent hours, not seats: $20 a month per unit of 100 hours, pooled across all five, with 15% off a year. ## What a team of five looks like with agents Five engineers, each running two or three agents, means ten to fifteen agents on the same codebase on a busy day, spread across Claude Code, Codex and Cursor. That is where the problems stop being occasional: - Agents on different laptops edit the same function and one overwrites the other. - A rename on one machine leaves stale call sites on four others. - Two people's agents solve the same problem in parallel. - Merges conflict in ways nobody saw coming, because the work was invisible until the branches met. Your existing practices still matter: an `AGENTS.md` everyone's agents read, small PRs, required status checks, code owners. Cork adds the live layer between machines that those practices do not have. ## Before you start: a 30-minute team checklist Do this once, with everyone in the room or on a call: 1. **Pick the shared repos.** Cork covers every repo on each machine, but agree which repos the team works on together. 2. **Check the agent instructions.** Make sure each shared repo has an `AGENTS.md` with build and test commands, conventions and no-go areas. Codex and Cursor read it, and Claude Code reads it when there is no `CLAUDE.md` (or through an `@AGENTS.md` import). 3. **Agree on landing.** Keep required status checks in CI. Cork's landing step complements them; it does not replace them. 4. **Name a coordinator.** One person starts the paid plan and answers questions for the first two weeks. ## Step 1: buy the team's agent hours The coordinator, on their own machine: ```bash curl -fsSL https://corkcli.com/install.sh | sh cork login cork upgrade ``` The install covers every repo on the machine, adds hooks to Claude Code, Codex and Cursor, and registers Cork's MCP tools for all three. `cork login` links the machine to an account. `cork upgrade` opens the workspace's billing page for units of agent hours; once they are paid for, the machine switches over by itself at its next prompt, with nothing to download. ## Step 2: invite the team, then everyone installs and logs in The coordinator invites the other four from the dashboard (Members, then Invite). Each of them accepts the invite, then, on their own machine: ```bash curl -fsSL https://corkcli.com/install.sh | sh cork login ``` Anyone with an agent session already open does not need to restart it: a session already open when Cork is installed is told at its next prompt. To confirm what Cork did on a machine: ```bash cork status ``` Once all five machines are on the team, agents across the team see each other on the shared repos. ## Step 3: add Cork to each shared repo's AGENTS.md In each shared repo, run: ```bash cork agents-md ``` This adds Cork's section to the repo's `AGENTS.md`. Commit it, so every agent on every machine reads the same short description of the protocol. ## Step 4: agree on five habits Cork's hooks do most of the work with no extra model calls: briefings, live awareness, the write gate and [semantic lint](https://corkcli.com/blog/semantic-lint) happen without anyone asking. A few habits get the rest of the value. 1. **Let the [write gate](https://corkcli.com/blog/write-gate) work.** When an agent's edit is held because another agent is changing the same code right now, the agent gets that teammate's version first. Do not route around it by switching files; let it adapt. 2. **Message the code, not the person.** `cork message "..."` messages the other agents in the repo. `--about path.py::symbol` reaches whoever is working on that function, so you do not need to know whose agent it is. `--to agent-xxxx` targets one agent. 3. **Record decisions with remember.** When an agent settles something others should know ("this endpoint must stay backwards compatible"), have it use `remember` with the code it applies to. Notes come back through `recall`, and the paid plan adds shared knowledge across the team. 4. **Check blast_radius before big changes.** Before a rename or signature change, ask the agent to run `blast_radius` on it. On a team, the callers it would break may be in someone else's active work. 5. **Read the recap.** `recap` shows what happened and what it cost. A short look each morning tells everyone what the team's agents did. ## Step 5: review after two weeks After two weeks, spend 20 minutes on: - How many merge conflicts and broken builds did you hit compared with a normal two weeks? - Which handoffs went through `cork message` and which still went through chat? - Did anyone's agent hit something the briefing should have covered? Cork publishes one benchmark you can compare against: in its benchmark (12 Claude Code agents at once on one repository, hard tasks), agents with Cork used 53% fewer tokens than the same agents without it. Your numbers will depend on your codebase and how you work. ## What it costs for five | Plan | Agent hours a month | Price | |---|---|---| | Free | 10 per account (30 bonus the first month) | $0 | | 1 unit | 100, pooled across all five | $20/month | | 5 units | 500, pooled across all five | $100/month | | 10 units | 1,000, pooled across all five | $200/month | What five people pay depends on how much their agents work, not on headcount. An agent hour is time the model is thinking and writing; a tool's run time (tests, builds), typing and idle time never count. Annual billing takes 15% off. Every member is a player, with teammates' machines, prompts matched to code by meaning (an embedding model), shared knowledge and 90-day history. Enterprise starts from about 100 units. If the plan ends, the workspace goes back to Cloud Free: the owner's agents stay coordinated, on the owner's own free agent hours, and the other four become viewers, whose agents stop getting briefings and collision checks. ## How Cork helps Cork is the [coordination layer](https://corkcli.com/blog/coordination-layer-ai-coding-agents) for multiplayer coding agents: it keeps Claude Code, Codex and Cursor agents, on one machine or across a team's machines, from overwriting each other on one codebase, and hands each agent what the others already learned. For five people, that means a rename on one laptop reaches every agent that depends on it on its next step, an edit against code another agent is changing right now is held until the agent has their version, and `git push` becomes one step that rebases onto teammates' commits, reruns the tests and pushes. Cork does not run agents, host machines or replace your CI and code review. Install on each machine with: ```bash curl -fsSL https://corkcli.com/install.sh | sh ``` ## Frequently asked questions ### How much does Cork cost for a team of five? Cork bills agent hours, not seats: $20 a month per unit of 100 hours, pooled across all five, graduated to $17 (units 11 to 50) and $15 (51 and up), with 15% off a year. You buy as many units as your agents use. ### Does every teammate need to use the same coding agent? No. Cork's hooks cover Claude Code, Codex and Cursor, and its MCP server works with any MCP client, so one person can use Codex while another uses Claude Code on the same repo. ### Do we still need branch protection and CI? Yes. Cork coordinates agents while they work and makes landing a rebase plus test run, but required status checks and code review remain your team's backstop. ### What if one teammate does not install Cork? Their agents will not see the team's agents, and the team's agents will not see theirs. Cork works best when everyone editing the shared repos has it installed. ## Sources - - - - ## Related reading - [From one machine to a team: moving from Cork Free to agent hours](https://corkcli.com/blog/cork-free-to-team): Move from Cork Free to agent hours: what changes when agents on teammates' machines see each other, how cork login and cork upgrade work, and what it costs. - [Cork Free: set up multiplayer coding agents on one machine in one command](https://corkcli.com/blog/cork-free-setup-one-machine): Set up Cork Free in one command so Claude Code, Codex and Cursor agents on one machine stop overwriting each other. No account; data stays on the machine. - [Messaging between your coding agents with cork message](https://corkcli.com/blog/cork-message-coding-agents): Use cork message to reach every coding agent in a repo, one agent, or whoever is working on a function, across Claude Code, Codex and Cursor. With examples. - [How to track AI coding agent costs per developer and per repo](https://corkcli.com/blog/track-ai-coding-agent-costs-per-developer-repo): Track Claude Code, Cursor and Codex spend per developer and per repository: /usage, OpenTelemetry with repo attributes, admin APIs, and tagging scripted runs. [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-team-of-five-setup