Conductor is the best way for one person to run Claude Code, Codex, and Cursor in parallel on a Mac, and since July it has shared cloud workspaces on its Pro tier. Coshell is the shared live session for your whole team, with as many agents and sessions as you want, on drives you host yourself. Here is the honest comparison.
Conductor's multiplayer runs in their cloud.
Coshell's runs on your own machines, free.
Overview is in with the three stat cards and the recent orders table. Live on http://localhost:5173.
Cards stretch now. The grid uses align-items: stretch and each delta line is pinned to the bottom with margin-top: auto.
Added the toggle. 30 days aggregates by week and the axis relabels. 8 tests pass.
Delta sits on the top row now, right aligned, so the numbers line up down the column.
Token bucket middleware is in with 12 tests. Step 1 verified.
Moved the bucket to Redis with a Lua script so the check is atomic across replicas. 14 passing. Steps 2 and 3 verified.
Webhooks now share the limiter. 16 passing. Marking step 4 verified and moving to the flag rollout.
Node.js isn’t installed on this VM, so I can’t start the dev server yet. I can install it with the system package manager and then run it. Want me to?
Understood, Node 22 LTS from NodeSource.
Server is running at http://localhost:5173/, Vite dev server ready in 284ms.
Eight commits since main: the coupon model, the checkout hook, and a migration. No PR yet.
No. You have no PLAN, TODO, or roadmap file.
Your closest thing to a plan is .coshell/review/dod.md: integer cents, handled promises, no focused tests, no dead exports.
Drafted docs/plans/discounts.md with five steps. It is waiting for approval in the Plan tab.
Hi. Big Pickle here, ready when you are.
2, 3, 5, 7, 11, 13, 17, 19, 23, 29.
1
2
3
4
5
6
7
8
9
10
11
12
bun test after every step, load test before the flag movesProtect /v2 from bursty API keys without slowing normal traffic. One key should never be able to starve the others.
A token bucket per API key, 100 requests per minute, stored in Redis so all three replicas share state. Over-limit requests get a 429 with a Retry-After header. Ship behind the rate_limit_v2 flag and roll out gradually.
bun test after every step. Load test with 1k keys before the flag moves to 50%. Watch 429 rate and p99 in the dashboard for a day at 10%.
Overview is in with the three stat cards and the recent orders table. Live on http://localhost:5173.
Cards stretch now. The grid uses align-items: stretch and each delta line is pinned to the bottom with margin-top: auto.
Added the toggle. 30 days aggregates by week and the axis relabels. 8 tests pass.
Delta sits on the top row now, right aligned, so the numbers line up down the column.
Token bucket middleware is in with 12 tests. Step 1 verified.
Moved the bucket to Redis with a Lua script so the check is atomic across replicas. 14 passing. Steps 2 and 3 verified.
Webhooks now share the limiter. 16 passing. Marking step 4 verified and moving to the flag rollout.
Node.js isn’t installed on this VM, so I can’t start the dev server yet. I can install it with the system package manager and then run it. Want me to?
Understood, Node 22 LTS from NodeSource.
Server is running at http://localhost:5173/, Vite dev server ready in 284ms.
Eight commits since main: the coupon model, the checkout hook, and a migration. No PR yet.
No. You have no PLAN, TODO, or roadmap file.
Your closest thing to a plan is .coshell/review/dod.md: integer cents, handled promises, no focused tests, no dead exports.
Drafted docs/plans/discounts.md with five steps. It is waiting for approval in the Plan tab.
Hi. Big Pickle here, ready when you are.
2, 3, 5, 7, 11, 13, 17, 19, 23, 29.
1
2
3
4
5
6
7
8
9
10
11
12
| Conductor | Coshell | |
|---|---|---|
| What it is | A Mac app that runs parallel coding agents in isolated workspaces, with a cloud tier | A shared live session your team runs its agents in, on drives you host yourself |
| Where multiplayer runs | Conductor Cloud, their microVMs, Pro and up | Your own machine, VM, or container, free |
| Agents and sessions | Several agents for one developer, one per worktree | As many agents and sessions as the team wants, shared or private, on one drive |
| Working the same branch together | One workspace per branch; a teammate forks a second branch | One working tree, queued visible prompts, no fork |
| Overlapping changes | Isolated until merge, then inspected by hand | Overlap layer catches the same concern being changed twice, while it happens |
| Context | Fresh agent and fresh checkout per workspace | Long-lived sessions, codebase loaded once per drive |
| Environment lifetime | Cloud workspaces sleep at 4h idle, stop at 23h50m | Drives persist; sleep when idle, wake in seconds |
| Clients | macOS app; API; SSH; mobile coming | Browser, CLI, VS Code, phone |
| Models | Claude Code, Codex, Cursor, OpenCode; your keys | 70+ models on your keys, including a Claude Pro or Max subscription |
| Privacy | Workspaces shared with the organization (per their docs) | Per session: private, named teammates, or whole drive |
| Git identity | Your local git and gh on your Mac | Commits by the prompter, PRs as the connected person, no token on the drive |
| Review | Diff viewer with line comments, Checks tab (ahead of ours) | Review tab in the session, plus your normal GitHub review |
| Self-hosting | Local Mac only; cloud is theirs | Your laptop, VM, or container, free; Enterprise on-prem |
| Pricing | Free local; Pro $50/mo up to 5; Teams $60/user/mo | Free to self-host; Enterprise is paid |
| Best for | One developer running many agents on a Mac | A team working together in live sessions, on its own machines |
Not for a solo developer on a Mac. Conductor is the better tool for one person running several agents in isolated worktrees, and its free tier is generous. Coshell is for the moment a second person needs to be in the same work: one live session the whole team is in, on a drive you host. Some teams will use both.
Conductor's multiplayer (July 2026) is a Pro feature that runs in Conductor Cloud: teammates open a shared cloud workspace, see who is there, and prompt together, at $50 a month for up to five people. Coshell's multiplayer is the product, and it is free on a drive you host: every prompt carries who sent it, sessions are private or shared, overlapping work is detected across sessions, and your code never leaves your machine.
Yes. A drive holds as many sessions as you want, each with its own agent, its own branch or worktree if you like, and its own privacy setting. The difference from Conductor is that teammates can be in the same session at the same time, and overlap between sessions is watched while it happens.
For file-level collisions, yes, and Conductor does it well. But isolation defers the other kind of collision, two changes to the same concern in different files, to merge time. Coshell's overlap layer looks for that while it happens, and a shared tree means there is no divergent copy to reconcile later.
Yes, on both. Conductor runs your Claude Code login locally. Coshell runs it on the drive through the Claude Code binary, alongside 70+ models on your own keys.
Conductor's local mode is your Mac, and its collaboration runs in Conductor Cloud. Coshell's self-hosted drives run on your laptop, a VM, or a container, free, and the whole team joins the same session in a browser; Enterprise adds org-wide on-prem and SSO.
Keep Conductor for solo work on your Mac. When the team needs to be in the same work, self-host a drive for free or open a hosted one, and put everyone in one live session.