dsh-sandbox-escalation-fix (DSH 0.1.6-alpha.1 supported, Win & Linux & macOS)

English | 简体中文

[!IMPORTANT] This is an independent community plugin. It is not published, maintained, or endorsed by DeepSeek, and it does not modify DeepSeek Harness core packages.

[!CAUTION] The DSH official team has made some improvements spanning from DSH 0.1.0-rc8 through 0.1.6-alpha.1, but still uses registry-global escalation schemas and execution-time validation. Users should first observe the built-in behavior and install this plugin only after reproducing the same-mode escalation, blank justification, or retry-loop failures described below.

Currently supported:
Latest supported DSH version: 0.1.6-alpha.1 (full list in Compatibility)
Desktop version: 2.0.3
OS: Windows & Linux & macOS (theoretically supported, not yet tested)

If there is any platform that has not yet been adapted and needs compatibility, please submit an issue.
If it's useful, please stars let more people can see it~ Thanks♪(･ω･)ﾉ

dsh-sandbox-escalation-fix is a zero-configuration compatibility plugin that directly resolves the issue of third-party models like GPT failing to call tools such as bash, pwsh, write, and edit under DSH All Access, resulting in repeated retries due to incorrect sandbox escalation parameter prompts.

If you've encountered the following errors, this plugin is designed for them:

Error: invalid justification: expected a non-empty sentence
Error: sandbox escalation to "danger-full-access" is not strictly wider than this call's current "danger-full-access" mode
Error: sandbox escalation to "workspace-write" is not strictly wider than this call's current "danger-full-access" mode

Contents

• What It Does
• The Problem It Solves
• Before and After
• Compatibility
• Quick Start & Installation
  • Release ZIP installation
    • Install into the default Web Profile
    • Install into another Profile
    • Build the Release ZIP
  • Command-line installation
  • Manual Windows Installation
• Upgrade & Maintenance
  • Upgrade an existing installation
    • GitHub commit installation
    • Manual Web Profile installation
• Uninstall
• Why This Plugin
• Verification, Behavior, and Plugin Cooperation
• Troubleshooting
• Contributors
• Development
• License

What It Does

This plugin makes DeepSeek Harness show the model only the sandbox escalation options that the current session can actually use.

In an All Access session (danger-full-access + never), the stock DSH tools still advertise sandbox_permissions and justification on bash, pwsh, write, and edit. But in that state:

• the session is already at the highest sandbox mode, so no wider mode exists;
• the approval policy is never, so every escalation request is rejected.

When a model fills in those parameters, the call fails before it runs. The model may then retry with different values and get stuck in a loop.

This plugin projects the model-visible tool schema per session, based on the live Sandbox Mode and Approval Policy. It also adds a minimal execution-time fallback: it handles redundant same-mode requests and, under workspace-write, removes a false danger-full-access request only from write / edit calls whose resolved filesystem boundary is inside the current workspace.

The Problem It Solves

• Models still see and send sandbox_permissions / justification in All Access sessions, so tools fail before they run.
• In workspace-write sessions, models see two escalation targets even though only danger-full-access is genuinely wider.
• With approval=never, models are still told escalation is possible when every request will be rejected.
• Tool descriptions and denial results keep saying ?escalation available,? which pushes the model to retry.
• Native Tool Call and PTC Mode SDK (formerly Code Mode) can show inconsistent capability surfaces.

Why It Happens

DSH tools expose static escalation fields when they are registered, while the modes that can actually be requested depend on each session's current Sandbox Mode and Approval Policy. The original model-visible schema is not projected from that live session state before the request is built, so a model can receive escalation parameters that cannot succeed. Tool validation then rejects those requests before execution, which can start a retry loop.

Before and After

Without the plugin, affected All Access sessions can repeatedly fail before the requested operation runs. The model alternates between an empty justification, a same-mode danger-full-access request, and even a downgrade request that DSH correctly rejects as not strictly wider.

Before: Repeated Validation and Escalation Errors

After: Tools Complete the Workflow

After installation, the same model can continue through Edit, Read, Pwsh, formatting, tests, lint, and type checking without entering the invalid escalation loop.

Compatibility

• Node.js ^22.19.0 or >=24.0.0
• @deepseek-ai/dsh-* 0.1.0-rc.5, 0.1.0-rc.6, 0.1.0-rc.7, 0.1.0-rc.8, 0.1.1-rc.1, 0.1.1-rc.2, 0.1.2-alpha.1, 0.1.2-alpha.2, 0.1.2-alpha.3, 0.1.2-alpha.4, 0.1.2-alpha.5, 0.1.2-rc.1, 0.1.3-alpha.1, 0.1.3-alpha.2, 0.1.5-alpha.1, 0.1.5-alpha.2, 0.1.5-rc.1, or 0.1.5-rc.2
• @deepseek-ai/cordis ^4.0.1
• Distribution: public npm Registry (package dsh-sandbox-escalation-fix, prereleases under the next dist-tag), GitHub Release ZIP, local .tgz, local directory, or a Git commit
• Operating systems: Windows (fully supported), Linux (verified on real Ubuntu 24.04), and macOS (expected compatible — the plugin is pure JavaScript and the .sh scripts are POSIX — but not yet tested on a real Mac). The plugin itself has no platform-specific code; actual sandbox enforcement on Linux/macOS depends on the sandbox backends available to the DSH host (Linux: bwrap or a Landlock-enforcing kernel 5.13+; macOS: Seatbelt), probed at runtime by DSH itself. When no backend is usable, DSH refuses to run the command rather than bypassing the sandbox. The plugin's permission projection and argument normalization do not depend on any particular backend.

The plugin checks the installed DSH package versions at startup. Mixed rc.5/rc.6/rc.7/rc.8/0.1.1-rc.1/0.1.1-rc.2/0.1.2-alpha.1/0.1.2-alpha.2/0.1.2-alpha.3/0.1.2-alpha.4/0.1.2-alpha.5/0.1.2-rc.1/0.1.3-alpha.1/0.1.3-alpha.2/0.1.5-alpha.1/0.1.5-alpha.2/0.1.5-rc.1/0.1.5-rc.2 installations and unknown DSH versions fail explicitly. An initially visible target with partial escalation fields or an incompatible output definition rejects that Agent's registration; a target that omits both escalation fields is accepted as already safe. During runtime, a Preset restriction or stable provider removal makes the wrapper dormant, while an incompatible replacement is isolated to that Agent and target tool and reported without terminating the Host process. A later compatible definition is wrapped automatically.

Quick Start & Installation

The plugin is a zero-configuration fix. The recommended path is to download the Release ZIP and install it into the Profile that runs the affected sessions. Start DSH normally after installation:

dsh --profile <profile>

You do not need to change the model configuration, Sandbox Mode, Approval Policy, or Agent Preset. The plugin projects the model-visible parameters from each Session's current permission state.

Install from the npm Registry (recommended)

The plugin is published to the public npm Registry as dsh-sandbox-escalation-fix. Current prereleases are published under the next dist-tag, while latest remains on the previous default release until a stable version is published. Install with an explicit @next or a pinned version so the selected channel is unambiguous.

dsh plugin --profile web add dsh-sandbox-escalation-fix@next

For another Profile, replace web:

dsh plugin --profile headless add dsh-sandbox-escalation-fix@next

To pin the exact version instead of following the tag:

dsh plugin --profile web add dsh-sandbox-escalation-fix@0.1.6-alpha1-win-linux

The npm package and the .tgz inside the GitHub Release ZIP come from the same build, so they behave identically. Restart DSH after installation.

Release ZIP installation

The 0.1.5-rc2-win-linux Release adds DSH 0.1.5-rc.2 compatibility. 0.1.5-rc.2 is byte-for-byte identical to 0.1.5-rc.1 across all 15 plugin-relevant published packages (Agent, LLM, Sandbox, Sandbox Policy, Scope, Session, Session Projection, System Prompt, Tools, Approval, Code Runtime, Tool Bash, Tool Pwsh, Tool FS, and Shell; 228 files each, 0 differences other than package.json, which only bumps each package dependency range), so the sandbox escalation schema and execution-time widening checks are untouched and no core plugin change is needed. This Release keeps DSH 0.1.5-alpha.1 support (Session V3, removal of ctx.agent, and the Inbox API changes do not affect this plugin) and passes all 40 tests and the TypeScript build against the complete public 0.1.5-rc.2 npm package set. It retains Windows and Ubuntu 24.04 Linux support, expected macOS compatibility, Desktop 2.0.3, linked/external plugin resolution, prepare-free Git installation, and the BOM-free configuration. Download dsh-sandbox-escalation-fix-0.1.5-rc2-win-linux-release.zip; it contains the tarball, install/uninstall scripts for Windows and POSIX shells, and the Chinese usage guide.

Close DSH before installing or upgrading the plugin. Ensure that dsh is available on PATH and that the running DSH version is rc5, rc6, rc7, rc8, 0.1.1-rc.1, 0.1.1-rc.2, 0.1.2-alpha.1, 0.1.2-alpha.2, 0.1.2-alpha.3, 0.1.2-alpha.4, 0.1.2-alpha.5, 0.1.2-rc.1, 0.1.3-alpha.1, 0.1.3-alpha.2, 0.1.5-alpha.1, 0.1.5-alpha.2, 0.1.5-rc.1, or 0.1.5-rc.2. Users should install only after reproducing the affected behavior.

Install into the default Web Profile

Windows:

powershell -NoProfile -ExecutionPolicy Bypass -File ".\install-release.ps1"

Linux/macOS (ZIP extraction does not preserve the executable bit, so invoke through sh):

sh ./install-release.sh

The script runs dsh plugin --profile web add <tgz-absolute-path>.

Install into another Profile

For example, use headless instead of the default web Profile. Windows:

powershell -NoProfile -ExecutionPolicy Bypass -File ".\install-release.ps1" -Profile headless

Linux/macOS passes the Profile name as the first argument:

sh ./install-release.sh headless

Build the Release ZIP

powershell -NoProfile -ExecutionPolicy Bypass -File ".\build-release.ps1"

The script builds lib, packages the npm tarball, then creates dsh-sandbox-escalation-fix-0.1.5-rc2-win-linux-release.zip in release/. The generated directory is ignored by Git; upload only this ZIP as the GitHub Release asset.

Command-line installation

Install into the exact Profile that runs the affected sessions and pin a reviewed commit SHA:

dsh plugin --profile <profile> add github:<owner>/dsh-sandbox-escalation-fix#<commit-sha>

Important: do not omit #<commit-sha>. Using only github:<owner>/dsh-sandbox-escalation-fix installs a snapshot of the repository HEAD at the time the command runs. Later repository updates are not synchronized automatically to the installed Profile, and the local snapshot cannot be assumed to match the repository's current code. Before diagnosing compatibility, verify the installed revision and explicitly upgrade to a newly reviewed commit SHA.

The package has no prepare or other install-time build script: pnpm installs the committed, prebuilt lib directly and no allowBuilds allowlist entry is needed. Versions before the prepare removal did run a build on install; if such an old version left a dsh-sandbox-escalation-fix@https://codeload.github.com/... entry under allowBuilds in $DSH_HOME/profiles/<profile>/pnpm-workspace.yaml, that stale entry can be removed after upgrading.

Run the installation command again, then inspect the composed configuration:

dsh --profile <profile> --dump-config

The output should contain a dsh-sandbox-escalation-fix bundle layer and the sandbox-escalation-fix plugin row. Start DSH normally after verification:

dsh --profile <profile>

Manual Windows Installation

A detailed Windows walkthrough ? Profile paths, folder layout, nested node_modules, and the correct replacement for an empty [] patch ? is available in README.zh.md.

For a compact file-by-file walkthrough, see Tutorials that even Peppa Pig can understand. The original Chinese layout is preserved in ???????????.txt.

The minimum manual layout is:

<profile-directory>\
??? cordis.patch.yml
??? node_modules\
    ??? dsh-sandbox-escalation-fix\
        ??? package.json
        ??? cordis.patch.yml
        ??? README.md
        ??? README.zh.md
        ??? lib\
            ??? index.mjs
            ??? index.d.mts
            ??? wrapper-protocol.mjs
            ??? wrapper-protocol.d.mts

Merge this block into the Profile's cordis.patch.yml; do not overwrite unrelated Profile patches:

- insert:
    - id: sandbox-escalation-fix
      name: dsh-sandbox-escalation-fix

Do not copy this repository's node_modules into the Profile. Multiple Cordis or DSH module instances can break Scope and Service identity.

To update an existing installation, follow Upgrade & Maintenance; do not repeat the Profile patch step.

Upgrade & Maintenance

Upgrade an existing installation

Close DSH before upgrading. The plugin package name, Bundle ID, and Profile patch row are unchanged, so an existing installation does not need another cordis.patch.yml entry.

npm Registry installation

Re-run the @next install command to pick up the latest prerelease:

dsh plugin --profile <profile> add dsh-sandbox-escalation-fix@next

If you previously pinned an exact version, replace the version in that command with the new one. Inspect --dump-config, then restart DSH.

GitHub commit installation

GitHub commit installations do not follow repository updates automatically. Even if main has fixed the issue or added support for a newer DSH version, the local Profile keeps its previously installed snapshot. Every upgrade must explicitly replace the old SHA with a new reviewed commit SHA and rerun the installation command.

Run the same installation command with the new reviewed commit SHA:

dsh plugin --profile <profile> add github:<owner>/dsh-sandbox-escalation-fix#<new-commit-sha>

This updates the Profile dependency. The repository ships prebuilt lib files and the package defines no install-time build scripts, so pnpm never asks for an allowBuilds entry. Inspect --dump-config, then restart DSH.

Manual Web Profile installation

Use the repository or packaged source that contains the new built lib directory. Open Windows PowerShell in that plugin directory and run:

powershell -NoProfile -ExecutionPolicy Bypass -File ".\deploy-web-profile.ps1"

The script uses $DSH_HOME when set, otherwise %USERPROFILE%\.dsh. It replaces only the eight published lib artifacts, compares every SHA-256 hash, and prints Deployment verified. only when the installed Web Profile exactly matches the new build. It does not modify the Profile patch or copy node_modules. Restart DSH after verification.

Uninstall

From an extracted Release ZIP, Windows:

powershell -NoProfile -ExecutionPolicy Bypass -File ".\uninstall-release.ps1"

Linux/macOS (Profile name as the first argument when needed):

sh ./uninstall-release.sh

Use -Profile headless (Windows) or headless (Linux/macOS) for another Profile. The equivalent CLI command is:

dsh plugin --profile <profile> remove dsh-sandbox-escalation-fix

The plugin lifecycle removes its wrapper hosts, wrapper layers, and result filters. Restart DSH, then confirm that --dump-config no longer lists the bundle.

Why This Plugin

It fixes the root cause, not just the error

Some fixes only delete the arguments after the model has already received a broken schema. Calls stop failing, but the model keeps seeing and sending the same unusable parameters.

This plugin projects the model-visible schema from the session's real permission state:

When the model cannot see a parameter that cannot succeed, it stops reaching for it.

It won't fix Native Tool Call but leave PTC Mode broken

Projection happens on the tool definition inside each Agent Exact Scope, so Native tool schemas and the PTC Mode SDK read the same result:

• Native tool schemas omit impossible escalation fields;
• PTC Mode TypeScript/Python SDKs omit them too;
• behavior stays identical across both modes.

It corrects false escalation only when existing access is sufficient

The execution fallback remains narrow:

• when requestedMode === effectiveMode, it treats the pair as an idempotent duplicate and removes it;
• when the session is workspace-write, the tool is write / edit, and both the lexical path and the nearest existing ancestor's real path remain inside the current Session workspace, it removes the model's false danger-full-access request and justification so DSH executes with the existing workspace-write policy.

Other requests remain untouched and go through normal DSH handling:

• danger-full-access + request danger-full-access — ignored, tool runs normally;
• danger-full-access + request workspace-write — not executed as Full access;
• read-only + a wider request — enters the normal approval flow;
• Shell escalation, paths outside the workspace, and paths that cannot be confirmed — keep their approval request.

This enhancement grants no additional access. If an apparently internal path crosses a symbolic link to an external target, DSH's existing filesystem sandbox still resolves the actual target and denies the out-of-workspace write; this plugin neither replaces nor bypasses that final boundary check.

It won't invent an approval reason

For genuine escalation requests, the plugin does not fill in a fake or placeholder justification. Missing, blank, or invalid reasons still go through DSH's own validation, so the approval flow sees honest, auditable input.

It won't say ?don't escalate? while results say ?escalation available?

When the session has no viable escalation target, the plugin also cleans up the natural-language side:

• the escalation guidance tail is removed from Shell tool descriptions;
• impossible escalation available hints are removed from Shell, filesystem, PTC Mode, and job_output results.

The model no longer receives contradictory instructions from the parameter schema, the description, and the failure output.

It won't modify or bypass DSH's security core

approveEscalation() keeps its strictly-wider check, approval flow, and one-shot authorization semantics. The plugin only projects the model-visible surface and performs the narrow execution-time normalization described above:

• no removal of the strictly-wider check;
• no auto-approval when approval=never;
• no extra permissions;
• no changes to DSH installation or core packages.

It won't treat every session the same

Wrapping happens per Agent/Session, never on the global tool registry. In the same process:

Session A = read-only + ask              ? sees two escalation targets
Session B = danger-full-access + never   ? sees no escalation fields

Each session is independent. If a session switches permission state mid-flight, the next model request gets a freshly projected schema.

It won't leave stale wrappers behind

The plugin listens to Agent creation, disposal, Preset changes, restrictions, and tool changes. When a dynamic Preset calls agent.ctx.tools.restrict(), the corresponding Exact Scope wrappers disappear synchronously with the restricted parent tools. Lifting the restriction restores the projected wrappers; tools absent during Agent creation are wrapped when they later become visible. Each Agent is coordinated independently, and disposed Agents or unloaded plugins restore the original definitions.

It won't lock you to a single DSH release

The plugin supports DSH 0.1.0-rc.5, 0.1.0-rc.6, 0.1.0-rc.7, 0.1.0-rc.8, 0.1.1-rc.1, 0.1.1-rc.2, 0.1.2-alpha.1, 0.1.2-alpha.2, 0.1.2-alpha.3, 0.1.2-alpha.4, 0.1.2-alpha.5, 0.1.2-rc.1, 0.1.3-alpha.1, 0.1.3-alpha.2, 0.1.5-alpha.1, 0.1.5-alpha.2, 0.1.5-rc.1, 0.1.5-rc.2, and 0.1.6-alpha.1. At startup it verifies that the installed @deepseek-ai/dsh-* packages are consistent and supported. Incompatible tool definitions fail explicitly instead of producing silent misbehavior.

It won't add configuration burden

Zero configuration. Install it into the Profile you actually use and start DSH as before. The 40-test suite runs directly against the complete public 0.1.5-rc.2 package set, and the 15 plugin-relevant packages of 0.1.5-rc.1 and 0.1.5-rc.2 have been compared file by file. It covers schema projection, false workspace-file escalation normalization, out-of-workspace and unconfirmed-root preservation, PTC Mode metadata, dynamic restrictions, multi-Agent isolation, delegate and wrapper-protocol replacement, internal timeout-budget forwarding, failure-hint cleanup, and unload behavior.

Compared with execution-only normalization

Verification, Behavior, and Plugin Cooperation

Behavior at a Glance

Verify the Fix

After installing or updating the plugin, fully restart DSH and create a new session.

A successful Web startup proves that the Profile composes and the plugin loads without a process-level failure. For a source checkout, this command starts the Web profile directly:

node --import tsx/esm apps/cli/src/bin.ts web

Run it with a supported Node.js version available on PATH. Wait for dsh web: http://127.0.0.1:3080, then perform the behavior checks below. Startup alone does not prove dynamic restriction behavior.

1. Select the previously affected OAI model.
2. Set Access Mode to All Access.
3. Ask the model to run pwsh and print the current directory.
4. Ask it to create a temporary file with write.
5. Ask it to update the file with edit, then read it back.
6. Switch workspaces and open existing sessions to confirm normal session restoration.
7. If the Preset uses agent.ctx.tools.restrict(), enter its restricted state and confirm hidden tools disappear; lift the restriction and confirm they return without recreating the Agent.

The calls should complete without sandbox_permissions argument errors or impossible escalation advice. Existing sessions should remain visible, workspace switching should work, and new sessions should be created in the selected workspace.

Verification Evidence

The 40-test suite uses real DSH SessionStore, ToolRuntime, AgentRegistry, SandboxPolicyService, ApprovalService, and SystemPrompt packages rather than only isolated mocks. It covers the permission matrix, schema projection, false workspace-file escalation normalization, parent-traversal and external-symlink preservation, PTC Mode SDK generation, dynamic restrictions, multi-Agent isolation, delegate replacement, cooperative wrappers, hint filtering, version checks, and unload behavior.

The runtime integration baseline is the complete public DSH 0.1.5-rc.2 package set, including SessionProjectionService. The official Session V3, Agent API, and Inbox API changes do not affect this plugin; 0.1.5-rc.2 is byte-for-byte identical to 0.1.5-rc.1 across all 15 plugin-relevant published packages except package.json, while Sandbox, Sandbox Policy, Approval, Bash, Pwsh, Session Projection, and Scope published artifacts are unchanged and the escalation schema contract is untouched. Automated tests do not completely replace E2E validation with a real model provider.

Wrapper Conflicts

The plugin owns the bash, pwsh, write, and edit names inside each Agent Exact Scope. Another plugin may share those names only through the explicit Symbol.for('dsh.tool-wrapper.v1') protocol. Cooperative layers are ordered by priority and owner.

An unknown same-name wrapper causes Agent registration to fail explicitly. In that case, remove one of the conflicting plugins rather than relying on an undefined load order.

Protocol types are exported from:

import {
  TOOL_WRAPPER_PROTOCOL,
  type WrapperLayer,
  type ToolWrapperProtocolV1,
} from 'dsh-sandbox-escalation-fix/wrapper-protocol'

Troubleshooting

Contributors

• sprainJinyu / 张金雨 — proposed the linked and external plugin package-resolution fallback in PR #5.
• tappat225 / tappat — removed the unnecessary Git dependency prepare step in PR #8, allowing installation without an allowBuilds entry.

Development

npm install
npm test
npm run build
npm pack --dry-run

Git, registry and tarball installations all ship the committed, prebuilt lib files; no build scripts run during installation, so pnpm never requires an allowBuilds entry. Run npm run build locally after changing src and commit the refreshed lib.

License

MIT