One shared live session where your whole team prompts the same coding agent, on the same files, terminal, and browser.
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
The agent made each person faster, on their own.
Coshell is the layer that puts the whole team in the session with it.
Not a transcript you share after. The live session you are in together, plus the layer that keeps overlapping work correct.
You and your teammates work in one session, live, watching the same work land on the same running environment. Presence shows who is here, and follow-along lets you watch a teammate steer the agent in real time. See how it differs from Claude Code.
One shared environment means the worst collisions never form, because there is no diverging copy to merge. But two agents can still reach for the same lines of the same file, and the engine underneath is last-write-wins. Coshell catches that: the second edit is held before it lands, and a coordinator steers that agent to proceed if the changes are compatible, or to take other lines, wait, or let the first land first. See how Overlap works.
Without a shared session, every teammate starts fresh and spends tokens loading the same repo again. In Coshell the context lives in the shared session and stays warm, so the team reuses the understanding one person already built, instead of paying to re-ingest the codebase for each developer. Reading from that warm context costs a fraction of loading it cold, and the more context-heavy the work, the more that compounds. Free to self-host, no seats.
Every session has its own privacy control. Keep a session private to just you, share it with a few named teammates, or open it to the whole drive. Experiment on your own, and open a session up only when you are ready.
Bring your provider keys and use any of more than 70 models. Admins pick the allowed set per drive and a fallback model, so the team stays consistent and spend stays predictable. Usage is billed by your provider, not by us.
The environment lives in the cloud, so the same session is reachable from any device. Open it in the browser, attach from your terminal with the CLI, or jump in from your phone to unblock a teammate. Same agent, same files, same running environment.
Need your own keys enforced org-wide, or Coshell running on your own infrastructure so your code never leaves it? Self-host does that for free; org-wide on-prem is the Enterprise tier.
Coshell is built on OpenCode, the open-source agent engine, and keeps it open underneath. The CLI, the SDK, and the wire protocol are unchanged, so the stock OpenCode CLI attaches to a Coshell drive as-is. What Coshell adds is the layer between the people: shared live sessions, overlap detection, per-session privacy, and free self-hosting for the whole team.
Tile sessions and terminals side by side, run several agents on one drive, and follow a teammate’s pane when you want to watch instead of type. Press Cmd+Shift+G on a Mac, Ctrl+Shift+G elsewhere.
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
Several people in the same live session at once, prompting one agent and watching its work as it happens, on a shared filesystem, terminal, and browser. It is closer to Figma or a Google Doc than to git: one source of truth everyone sees live, instead of each person taking a copy and merging later.
Coshell works with those agents rather than replacing them, across more than 70 models. The single-player tools made each individual faster while leaving 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 standardized environment. See Coshell vs Claude Code and Coshell vs Codex.
No, for two reasons. One shared environment means there is no diverging copy to lose, so the worst class of collision cannot form. And when two agents reach for the same lines of the same file, an overlap layer catches it and holds the second edit before it lands, then steers that agent to proceed if the changes are compatible, or to take other lines, wait, or let the first land first.
Every session has its own privacy control. Keep a session private to just you, share it with a few named teammates, or open it to everyone on the drive. Experiment on your own, and open a session up only when you are ready.
A cloud development 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 overlap detection and shared agent context. The difference is the shared, agent-native, multiplayer design, well beyond simply being hosted in the cloud. See Coshell vs Codespaces.
We do not become your source of record. Work in a drive commits and pushes to the repositories you already have, so your branching, review, and CI carry on unchanged, and there is nothing to migrate off if you leave. The Enterprise tier adds org-wide on-prem and SSO; either way, self-host keeps your code on your own machine.
No. A drive is a persistent, warm session, so multiplayer is real-time when you want it and a clean handoff when you do not. A teammate in another timezone opens the same drive and picks up exactly where you left off, with the agent's context intact, instead of cold-starting. For a team working together it is live co-working; for an async team it is a handoff with all the context preserved.
It works alongside them. Most engineers keep their own editor and agent for solo work and drop into a shared drive when the work is shared: pairing, onboarding, live debugging, review. Coshell is built on OpenCode, so your stock OpenCode CLI attaches to a drive unchanged, and work pushes to your normal git. Adopting it is additive, not a migration. Coshell is the shared session your team works in, not another editor to switch to.
A model owner is built to make one agent better. The hard part for a team is coordination across people and across models, and that layer is only useful if it stays neutral: many models on your own keys, your own git, your own infrastructure. A model owner is disincentivized from building exactly that, because neutrality works against their own model. The neutral layer that lets any agent coordinate with any other is something only a company that does not own a model is positioned to build.
Open a drive in the browser, or attach from the CLI. Your whole team works in shared live sessions, on the same codebase.