Huanqi Cao
d8acd2b656
The self-hosted Linux runners ship no pwsh, and the pwshAvailable probes used spawnSync('where.exe'), which reports a missing binary as status null instead of throwing — the suites never skipped and failed with spawn pwsh ENOENT. Probe with resolvePwshPath() status instead, the same gate the coverage exemption uses. Exempt pwsh-sandbox src from coverage on pwsh-less hosts (its remaining helpers branch and the invariant companion ride the executor suites' real pwsh runs); pwsh-ful hosts keep the full 100% bar. Cover the windows-acl probe case and the runner-entry resolution in sandbox-local on Linux (chain-seam tests plus a windowsAclRunnerEntry seam) — the package's POSIX-only suites are Linux's only chance to cover the new lines. Static gate fixes: declare dsh-pwsh-sandbox in the base bundle, register the runner files entry in constraints, knip entries for the e2e suite and where.exe, regenerate the module graph. Verified in WSL (no-pwsh Linux): pwsh-sandbox 5 pass/13 skip with the exemption active, sandbox-local coverage 100%.
@deepseek-ai/dsh-pwsh-sandbox
English | 中文
Sandbox-consuming PowerShell implementation of the ctx.bash executor seam: every command runs as pwsh -NoLogo -NoProfile -NonInteractive -Command <command> confined through ctx.sandbox, with the selected mode, enforcement, and denial facts stamped on each settled result. The pwsh twin of @deepseek-ai/dsh-bash-sandbox, a call-for-call mirror per the pwsh executor and tool decision — the confinement substance is platform-neutral: on Windows the sandbox seam resolves to the ACL restricted-token runner chain (@deepseek-ai/dsh-sandbox-windows-acl), on Linux/macOS to bwrap/Landlock/Seatbelt.
The executor inherits @deepseek-ai/dsh-pwsh-local's process mechanics and consumes its argv-level seam (argv() / runArgv() / startArgv() / onProcessDone()) to wrap the exact pwsh invocation through the provider. The sandbox policy (mode + workspace root) is NOT this package's config: it rides each call from ctx.sandboxPolicy (tool calls pass the calling session's resolved policy; direct calls fall back to deployment policy).
Behavior
danger-full-access: commands run through the local executor unchanged; results carrysandbox: { mode, denied: false }.- Confined modes (
read-only,workspace-write): the pwsh argv is wrapped byctx.sandbox.confine(); runner-launch refusal fails closed withSANDBOX_UNAVAILABLE(foreground throw, backgroundrunnerFailedfact), and a denied write classifies against the selected backend'sdenialSignaturesintosandbox.denied.
Model Experience
Confinement works, denial surfaces as command failure
What the model sees
The confined command's own stderr (e.g. Access to the path '...' is denied. under the Windows ACL runner); the tool layer converts classified denials into the standard permission-denied surface exactly as it does for the bash tool.
Token effect
No model-visible text beyond the command's stderr and the tool layer's standard denial surface.
KV Cache effect
None directly; the denial surface belongs to the tool layer.
Known Limitations and Deferred Work
- Reads are unrestricted on Windows (the ACL runner restricts writes only); the read boundary is documented in
@deepseek-ai/dsh-sandbox-windows-acl. - The Windows workspace-write temp area is the real temp directory (
GetTempPathW). This is a deliberate backend-defined choice, the same decision Landlock makes (readWrite: ['/tmp', ...]): the seam's "backend-defined temp area" permits it, and the escape probe intests/acl.e2e.tslives outside the temp tree for exactly that reason. A per-run private temp (bwrap's--tmpfs /tmpsemantics) would additionally need an environment-block rewrite in the runner; it is an optional future hardening, not a correctness gap. - Windows read-only is strict zero-grant — not even the NUL device is writable;
> $nullredirection still works (documented in the backend package).