DSH Marketplace
中文

dsh-auto-review

PerryLink/dsh-auto-review

Second-model auto-review on the approval answerer chain: a read-only reviewer subagent returns structured allow/deny verdicts with reasons, fail-closed by default.

761TypeScriptApache-2.0Source

Install

Add dsh-auto-review to DeepSeek Harness

via npm

Resolves a published tarball rather than cloning the repository, and installs without any extra setup. Swap `web` for your profile name if you run another one.

via GitHub · npm package

Installing from GitHub runs the project's build script, which pnpm blocks until you allowlist it — run the command once and pnpm prints the exact key to add under `allowBuilds` in ~/.dsh/profiles/web/pnpm-workspace.yaml.

What happened when we ran it

Installed cleanly when we ran it

Every command here is run in a throwaway container against a clean profile, and the result is whatever the harness recorded — not a guess from the source. Last run 6d ago.

Due diligence

Before you install dsh-auto-review

  • Source of record: PerryLink/dsh-auto-review — present in the community registry that DSH's own plugin market installs from.
  • Licensed under Apache-2.0.
  • Detected: terminal surface, requires credentials. Read the source before granting these.
  • A listing here is not a security review. Plugins run with your agent's permissions.

The AI take

What it is — DeepSeek Harness plugin providing second-model auto-review on the approval chain.

Who it is for — When you use an agent in DeepSeek Harness to handle sandbox boundary actions, needing structured allow/deny review results. If your workflow relies more on human approval rather than subagent decisions, you don't need to install this plugin.

Watch out — Sandbox test passed, successfully installed in new profile and registered by harness. Requires providing API key or token upon installation. Will execute shell commands.

The verdict — I would install it because it implements a fail-closed default policy that prevents unsafe actions from passing when review fails.

Generated by grok-4.6, and a starting point rather than a verdict. Where it says a plugin installs or does not, that is from a real run in a clean profile — everything else is read off the repository. Trust the source over this.

What dsh-auto-review does

Abstract diagram of evidence flowing through a read-only review gate into an allow path or deny barrier.

dsh-auto-review is a DeepSeek Harness plugin that adds a second-model, read-only review step to approval requests before actions cross the sandbox boundary. DeepSeek Harness loads an approval/request answerer that claims requests covered by the ai policy and delegates other requests with next(), preserving the existing human approval chain. For an AI-reviewed request, a one-shot reviewer subagent reads the workspace with read, glob and grep, the streamed tool-call arguments with sensitive values redacted, the request reason and configured risk rules. It returns { decision, reason, riskLevel }; the reason is linked to the denied tool result by callId.

The default fallbackPolicy is rejected, so a reviewer crash, timeout or schema mismatch fails closed. autoReview/verdict and autoReview/rejection events provide a session audit trail, while /auto-review on, /auto-review off and /auto-review approve control session behaviour. Policies and limits are configured in cordis.yml, including toolsPolicy, riskRules, reviewerModel, reviewerTimeoutMs and reviewerTools.

dsh-auto-review suits workflows that need model-based evidence review without giving the reviewer write or terminal tools. It is the wrong choice when every request should remain a human decision, or when a second model and its credentials should not inspect workspace evidence. The listed risk surface includes a terminal surface and required credentials; the shipped policy reviews bash and write with AI, while other tools, including edit, remain on the human chain unless configured otherwise.

dsh-auto-review documentation

How dsh-auto-review behaves

DeepSeek Harness invokes the plugin on the approval/request answerer chain. The plugin recognises reviewer requests by identity and delegates them, while the reviewer child is bounded by maxDepth and its non-empty reviewerTools allow-list. Each decision follows approval/asked to autoReview/verdict or autoReview/rejection, then approval/decided. Audit events are log-only and marked ignorable: true. An optional invariant companion checks that rejection markers and events correspond.

A deny reason is injected into the calling model's denied tool result. Fallback and never decisions add [auto-review-fallback] and [auto-review-never] markers. The rejection circuit breaker acts after three consecutive denials, or six of the last ten verdicts, per turn.

Configuring dsh-auto-review

DeepSeek Harness reads these Schemastery fields from cordis.yml. An id-targeted override replaces the complete row, so required keys must be restated.

Key Default Purpose
enableByDefault true Initial session state; /auto-review on|off creates a durable override
toolsPolicy.default human Policy for unlisted tools
toolsPolicy.overrides {} Per-tool ai, human or never policy
riskRules [] Regex rules for reason, toolName or arguments
reviewerModel inherit Reviewer model route
reviewerTimeoutMs 60000 Verdict deadline
fallbackPolicy rejected Failure result: rejected, delegate or allow-once
maxReviewsPerTurn 10 AI-verdict budget
maxFailuresPerTurn 10 Reviewer-failure budget

Other documented fields include reviewerProvider (fork), reviewerTools ([read, glob, grep]), reasonMaxChars (2000), reviewerGuidance, reviewerPolicyText and denyGuidance.

Commands and requirements

DeepSeek Harness provides /auto-review on, /auto-review off and /auto-review approve. The last is a one-shot override. The plugin targets DeepSeek Harness 0.1.1-rc.2, with peers >=0.1.0-rc.8 <0.2.0, and requires Node ^22.19.0 || >=24.0.0. It supports all platforms as a host answerer; the optional Web review panel uses the session-projection capability. The reviewer inherits the session agent's route unless reviewerModel is set.

Written from the project's own documentation and kept in sync with it. Where the two disagree, the source is authoritative — read the README on GitHub

Same category

Alternatives to dsh-auto-review

dsh-agent-teams

NanmiCoder

829

AgentTeams multi-agent teams.

Installed cleanly when we ran it

npm packageTypeScriptyesterday

AI review

dsh-agent-teams

What it is — Converts a DeepSeek Harness session into a multi-agent team with role assignment, dependency-aware tasks, and direct messaging.

Who it is for — When I need to break a complex goal into dependency-aware subtasks and coordinate multiple continuable sub-agents, I would use this plugin. Users who only need single-agent execution would not require it, as the plugin requires binding work to a single captain-led team.

Watch out — Sandbox testing confirmed successful registration and enabling in a fresh profile. No obvious issues were found. Team state is persisted in a file-backed directory, with no coordination for concurrent process edits to the same team.

The verdict — I would install it for persistent multi-agent coordination with automatic task claiming, but only if staying within single-team limits to avoid issues with concurrent process access.

Generated by grok-4.6, and a starting point rather than a verdict. Where it says a plugin installs or does not, that is from a real run in a clean profile — everything else is read off the repository. Trust the source over this.Read the source
1Details

dsh_workflow

icetomoyo

97

UltraCode-style multi-agent orchestration: a generatable, savable, governable, observable, resumable workflow layer.

Installed cleanly when we ran it

GitHub sourceTypeScript10d ago

AI review

dsh_workflow

What it is — This plugin upgrades DeepSeek Harness one-time multi-agent scheduling into a generatable, savable, resumable workflow layer.

Who it is for — When running repeated complex multi-agent tasks like parallel investigations in DSH sessions, this plugin is suitable for saving workflows as reusable capsules. If your tasks are only brief one-time scheduling, this plugin's workflow layer is unnecessary because it focuses on higher-level flow product capabilities.

Watch out — Source installation requires manual approval of the build script. Sandbox test passed in a fresh profile and registered successfully. No obvious issues found.

The verdict — I would install it because it turns multi-agent into a long-term maintained agent workflow library.

Generated by grok-4.6, and a starting point rather than a verdict. Where it says a plugin installs or does not, that is from a real run in a clean profile — everything else is read off the repository. Trust the source over this.Read the source
1Details

dsh-tabbit

Tabbit-Browser

92

Tabbit Browser plugins for Deepseek Harness

Installed cleanly when we ran it

npm packageJavaScript2d ago

AI review

dsh-tabbit

What it is — dsh-tabbit plugin integrates Tabbit Browser for DeepSeek Harness to provide tabbit-browser browser automation skills.

Who it is for — It is suitable when the agent needs web automation, information extraction or QA. Users who are already running Tabbit Browser locally can choose not to install it.

Watch out — Sandbox testing passed: installed in a new profile and registered in DeepSeek Harness. No obvious issues found.

The verdict — I would install it because it allows the agent to directly control Tabbit Browser via tabbit-cli.

Generated by grok-4.6, and a starting point rather than a verdict. Where it says a plugin installs or does not, that is from a real run in a clean profile — everything else is read off the repository. Trust the source over this.Read the source
1Source

dsh-plugin

Tabbit-Browser

83

Tabbit Broser plugins for Deepseek Harness

Installed cleanly when we ran it

GitHub sourceJavaScript6d ago

AI review

dsh-plugin

What it is — The plugin provides Tabbit browser control capability for DeepSeek Harness.

Who it is for — When performing automation tasks with DeepSeek Harness that require browser integration, this plugin enables Tabbit control. It is not necessary if your DSH runs in a sandbox without local GUI Browser access, because this Skill does not make Tabbit automation available.

Watch out — Sandbox testing passed: installed in fresh profile and registered by harness. Static check indicates it can execute shell commands. From source install, the build script needs manual allowance first. Requires Tabbit browser version 1.9.0 or higher.

The verdict — I would install it because it handles Tabbit installation and Skill loading via background tasks, but only if local Tabbit GUI access is possible.

Generated by grok-4.6, and a starting point rather than a verdict. Where it says a plugin installs or does not, that is from a real run in a clean profile — everything else is read off the repository. Trust the source over this.Read the source
1Source

mstar-harness

btspoony

53

Skill-driven harness/loop engineering workflow agent plugin.

Installed cleanly when we ran it

GitHub sourceTypeScripttoday

AI review

mstar-harness

What it is — Morning Star Harness is a TypeScript Harness Workflow Engine Agent Plugin that enforces deterministic workflow gates and provides mstar-* judgment skills.

Who it is for — When performing iteration or codebase audit in DeepSeek Harness for multi-agent code delivery, it fits. If your tasks do not involve judgment skills for engineering loops, you don't need to install it.

Watch out — From source install you need to manually whitelist the build script. It automatically executes scripts and runs shell commands. Sandbox test passed: installed in new profile and registered by harness, no obvious issues found.

The verdict — I would install it because it provides a portable engine across multiple hosts.

Generated by grok-4.6, and a starting point rather than a verdict. Where it says a plugin installs or does not, that is from a real run in a clean profile — everything else is read off the repository. Trust the source over this.Read the source
2Details

dsh-automation

titanwings

51

Scheduled coding runs in fresh agent sessions with auditable history.

Installed cleanly when we ran it

GitHub sourceTypeScript6d ago

AI review

dsh-automation

What it is — dsh-automation runs self-contained coding tasks on schedule in fresh agent sessions, leaving an auditable run history.

Who it is for — If you need to run recurring or one-shot coding work later without relying on old chat, each in an explicit workspace with permission boundary and inspect the history, dsh-automation fits. For tasks that depend on interactive approval midway through or react to file, HTTP, or process conditions rather than time, users do not need it as it targets time-based standalone executions.

Watch out — Installation from source requires Node.js 22.19 or newer. Sandbox test passed in fresh profile with harness registration. No obvious issues found.

The verdict — I would install it because sandbox tests confirm it runs in fresh sessions with auditable history for standalone coding tasks.

Generated by grok-4.6, and a starting point rather than a verdict. Where it says a plugin installs or does not, that is from a real run in a clean profile — everything else is read off the repository. Trust the source over this.Read the source
1Details