Stop computer bash commands on the computer itself

The sandbox ignores sandbox_execute_bash's timeout and runs every call in
one shared shell session, so a command that never ended (a network read
without a timeout, a prompt) kept running after the call gave up, and
every later command on that computer, from any run, timed out behind it.
Each command now runs in its own bash under coreutils timeout (at most
ComputerCommandMaxSeconds, below McpToolTimeoutMs), reading no input, and
says when it was stopped.

Replies' env list says where each key comes from and that workspace,
computer and install-wide envs reach every bot there, so bots stop
telling users a workspace's keys must be set per bot. The computer
prompt says installed programs may be gone after a restart. An empty
open-only task list says closed tasks aren't shown, and update_task says
when a status was already set.

Co-Authored-By: Claude Opus 5.5 <[email protected]>
This commit is contained in:
2026-10-28 04:34:39 +01:00
co-authored by Claude Opus 5.5
parent e51f2aa9a4
commit 2896e76256
10 changed files with 165 additions and 12 deletions
+2 -1
View File
@@ -137,7 +137,8 @@ A worker reaches another worker **only through the coordinator**: `ask_crew` (ev
- **Browser profile:** `DATA_DIR/computer-browsers/<id>` mounted at `~/.cobble-browser`, which Chrome is pointed at (`BROWSER_EXTRA_ARGS=--user-data-dir=... --password-store=basic`), so logins, extensions and site permissions survive restarts and recreates. Never use the image's `~/.config/browser`: every boot overwrites its `Default/Preferences` with session-only cookies. Seed prefs (`computer-browser-preferences.ts`) are written only when missing; stale `Singleton*` links are removed before each provision. Bumping `AppData.ComputerWorkspaceLayout` recreates every computer once (`migrateComputerLayout`, skipped while a job runs on it). Default extensions (`ComputerBrowserExtensions`, `normal_installed`: can be turned off, not removed) come from a managed policy written as root into the running container (`syncComputerBrowserPolicy`, on provision and every `sync-computers` run), not a mount, so existing computers get new ones without a recreate; Chrome installs them live. Deleting a computer keeps its directory. Outside docker (dev) containers are reached by IP (`pickReachableAddress`).
- One main computer (`ensureMainComputer`), whose coordinator grant is locked (`isLockedComputerAccess`). With `MAIN_COMPUTER_URL` set (dev) it isn't provisioned. Only super admins create/delete/retry computers.
- A bot's effective computers (`grab-bot-computers`): the workspace's in a workspace; otherwise direct grants plus its workspaces' (coordinator: all); `bots.computer_id` first. Only a `running` (healthy) computer is reached (`grabComputerMcpUrl` `ready`, `computerNotReadyMessage`).
- **Folders (enforced in tools, not a Linux user):** a run works only in `bots/<bot>/users/<user>` (cwd, `HOME`), `bots/<bot>` and `workspaces/<workspace>` (`computerRunDirs`, `withComputerDirs`, `resolveComputerPath`; `sudo` refused).
- **Folders (enforced in tools, not a Linux user):** a run works only in `bots/<bot>/users/<user>` (cwd, `HOME`), `bots/<bot>` and `workspaces/<workspace>` (`computerRunDirs`, `withComputerDirs`, `resolveComputerPath`; the file tools' `sudo` refused; bash runs as `gem`, who has sudo, so packages can be installed, and they're lost on a restart).
- **Bash on a computer:** the sandbox ignores `sandbox_execute_bash`'s `timeout` and runs every call in one shared shell session, so a command that never ends blocked every later one on that computer, from any run, and Stop never killed it. `withCommandTimeout` runs each in its own `bash -c` under coreutils `timeout` (at most `AppData.ComputerCommandMaxSeconds`, below `McpToolTimeoutMs`), stdin from `/dev/null`. Never `exit` in a wrapper (it would kill the session shell).
- **Viewing:** `/computers/<id>/` proxy (`handleComputerProxy`, strips app cookies, bridges websockets). `ComputerView` modes: browser, terminal, code, desktop (noVNC; the only one with Chrome's toolbar). **Take-over:** `ask_user_to_take_over` writes `[take-over:ID:mode]`; "Hand back" sends a chat message; tasks stop and report rather than wait.
### Schedules (routines in the UI)