> ## Documentation Index
> Fetch the complete documentation index at: https://docs.bead.run/llms.txt
> Use this file to discover all available pages before exploring further.

# Sandbox data storage

> Follow sandbox data storage across source files, outputs, environment values, and snapshots to find the right cleanup operation.

Plan cleanup for each resource you created. Sandbox data storage has separate lifetimes for machines, saved snapshots, vault values, and local caches.

## Source sync

[Worktrees](https://docs.bead.run/worktrees/sync.md) send tracked and nonignored untracked source inputs to `/workspace` on the remote machine. Safe explicit includes can add ignored inputs. Sync excludes raw Git data, secret paths, control files, and build-output directories.

The remote workspace has files and an initialized Git repository, without the laptop's Git history. Guest-only caches and build outputs can survive later syncs. Exact managed source deletes propagate.

The CLI also keeps private frozen source bytes in its local source cache. Removing a remote machine does not remove that local cache.

## Guest files and command output

Uploaded files and command-generated files live in the machine's disk and memory. Artifact pulls write into a new private local cache directory outside authored source.

Worktree execution output can be recovered by execution ID. Its spool is bounded to 64 MiB and remains until the next managed command or machine deletion. See [Recover execution and clean up](https://docs.bead.run/worktrees/recovery.md).

## Environment values and profiles

Worktree laptop env values enter remote process environments. Imported project values are stored in the encrypted project vault and delivered according to project and service scope. `bead env ls` lists names only.

Credential profiles and human subscription sign-ins use their own vault and delivery paths. See [AI credentials and vault](https://docs.bead.run/agents/credentials.md).

## Snapshots and forks

Snapshots capture memory, disk, and device state. This can include process environment values, browser sessions, files, and cached application identity. Fork children inherit captured contents. Deleting the source machine does not delete separately retained snapshots or independent children.

## Remove data by resource

Delete finished machines, retire saved snapshots, remove unneeded vault names, and clear local caches separately. Financial purchases and receipts remain audit records. See [Data retention and deletion](https://docs.bead.run/security/retention.md).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.