Your team doesn’t hand off code anymore, it hands off context. Coshell hosts live sessions your whole team works in, with the same files, the same memory, and the same running app. Nobody resumes from your notes. They resume from the session.
Start a driveOverview 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
AI coding tools made developers individually multiplayer with machines; Coshell makes the team multiplayer with the machine.
Coding agents are single-player. They made one engineer faster and left the team’s coordination worse. Context gets duplicated, work happens where nobody else can see it, and a handoff means pasting memory files to the next person.
Coshell is a shared environment for people and agents. Several people prompt the same live agent inside one drive with a filesystem, a command environment, and a localhost browser. Your teammate doesn’t resume from your notes. They resume from the session.
Closer to a shared document than to git. Instead of everyone taking a copy and merging later, there is one running environment that everyone sees live.
Prompt together Everyone prompts the same agent. Prompts queue in the open, with a name on each one, so nobody wonders what it is doing next.
Hand off nothing A teammate in another timezone resumes the session itself, with the agent’s memory and the running environment intact.
Follow along See who is on the drive and what they are looking at. Follow a teammate and your view tracks theirs: the session, the file, even their cursor in the preview.
Every dev server in the drive shows up as a Browser tab. Teammates see the same page and each other’s cursor. Turn on Inspect, pin a comment on any element, and send the pins to the agent as one prompt.
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%.
You’re not writing the code anymore, you’re steering the work, so the real work is the plan. A plan is approved by the team, executed step by step with each step verified, and anyone can comment on a line of it while the agent is working.
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.
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%.
Before an edit lands, Coshell checks whether another session is working the same path. If it is, the edit is held and a reasoning arbiter reads what both sides are actually changing, then tells that agent how to proceed. Every decision is recorded. It fails open, so it never stalls work.
A drive runs wherever you want it to: the laptop you are typing on, the Mac mini under the desk, one click on Render, Railway, or Northflank, or a cloud-init file on any VM. It binds to localhost, connects out over a WebSocket, and opens no inbound ports, so your code never leaves your machine.
$ curl -fsSL https://coshell.ai/install | bash $ coshell host initYour own laptop is a drive too. Teammates reach it through the tunnel while it is awake, and the doctor warns you before the machine would sleep on them.Waiting for the first command
$ curl -fsSL https://coshell.ai/install | bash $ coshell host initThe Mac under the desk becomes a drive for the whole team. A launchd agent keeps it running and reconnects after a reboot.Waiting for the first command
#cloud-config
runcmd:
- curl -fsSL https://coshell.ai/install | bash
- coshell host initDigitalOcean, Hetzner, Vultr, Linode, or anything that takes cloud-init. Paste one file into the machine’s user data.Waiting for the first commandEvery session has its own chat. Mention a teammate to pull them in, mention a file to point at it, ask @coshell what the team is up to, follow someone to see what they see, and get notified in the web app or the terminal.
Grid mode tiles as many sessions as you want in one window. Five agents on two drives, a terminal under any of them, and a teammate’s pane you can follow when you want to watch instead of type.
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
Coshell sits on top of the agents and the git workflow you have. Nothing to migrate.
Your agents Claude Code, OpenAI, and more than 70 models. Admins choose which providers the team can use and set a default model, applied live to every running drive.
Your git Work in a drive pushes to your repository as a normal git operation. Branching, review, and CI stay exactly as they are.
Your terminal coshell attach opens the same drive in a TUI, with presence, chat, and follow, so nobody has to leave the terminal.
Any device The environment lives in the cloud, so a session is reachable from a laptop or a phone. Teams have fixed live issues from a phone.
Three ways in. They all end in the same drive, with your team on the link.
coshell signup. It creates the account, names the org, and enrolls this machine as your first drive, in one go.Already signed up? coshell host init adds this machine as another drive.
Coshell works with those agents rather than replacing them, across more than 70 models. The single-player tools made each engineer faster and left the team’s coordination worse. Coshell is the layer that makes the team work with the agent together: shared live sessions, overlap handling when code is written, shared context, and one standard environment.
A cloud dev environment gives one person a machine in the cloud. Coshell is built around several people and the agent working in the same live environment at once, with prompts that queue in the open, overlap negotiation, and agent context that everyone shares.
Because everyone works in one live environment instead of separate copies, the worst kind of collision never forms. When two efforts still touch the same concern, the overlap layer detects it, holds the second edit, and a reasoning arbiter decides how that agent should proceed based on what both sides are actually changing. It never merges code, and every decision is recorded.
No. Most engineers already sit in a terminal in front of a coding agent. Coshell gives you the same session in the web app or the TUI, your git workflow stays exactly as it is, and a new teammate gets a working environment in seconds instead of a day of setup.
On the drive, which can be your own machine: a Render or Railway worker, any VM, a Mac mini, or a laptop. The drive connects out to the edge over a WebSocket and opens no inbound ports. Your repositories stay yours through normal git, so there is nothing to migrate away from.
Free to start on your own machine: you bring the compute and nothing is metered. Enterprise adds SSO, role-based access, an audit trail, priority support, and Coshell on your own infrastructure at scale. Email enterprise@coshell.ai and we set your team up.
Free to start on your own machine. Enterprise when you need SSO, an audit trail, or on-prem at scale.