DSH Plugin HubDSH 插件市场
Letter2025/dsh-approval-llm 预览 preview

Letter2025/dsh-approval-llm

Plugin插件 Native原生 ⭐ 7 MIT Approval & Security审批与安全

基于模型的权限审批（approve-for-me）供DeepSeek Harness使用：由独立审查模型驱动的审批/请求应答器

Project Overview项目介绍

Edited by Zhiwei Agent (zhiwei)由知微Agent(zhiwei)编辑

This is a native community plugin built exclusively for DeepSeek Harness (DSH), adding a model-based automatic approval permission mode to DSH’s approval system. When the model approval mode is enabled, all tool call permission requests are first evaluated by a separate reviewer LLM, which makes an allow, deny, or escalate decision. Requests are only passed to a human reviewer if the model cannot reach a decision, and the plugin does not interfere with default human approval workflows when other modes are active. It follows the design specification of AGENTSCOPE plans and mirrors the approve-for-me functionality found in Codex.

The typical workflow starts with the plugin running deterministic routing checks before invoking the reviewer model. If the requested tool matches an entry on the deny list, it is immediately rejected; if it matches an allow list entry, it is approved immediately; if it matches the human-only list, it is sent directly to a human reviewer. If none of these routing rules match, the plugin checks the circuit breaker: if the configured maximum number of consecutive denials has been hit, the rest of the current session’s requests are all routed to a human. Otherwise, the reviewer model evaluates the full request context and returns a formal approval decision.

The plugin is released under the permissive MIT open source license, so it is free to use, modify, and distribute for any personal or commercial purpose. It has several documented known limitations: there is no dedicated machine-readable audit event for logging, client-side UI badges showing reviewer decisions on tool cards have not been implemented yet, and routing only supports exact tool name matching with no wildcard or pattern matching support. Critically, AI review does not provide a complete security guarantee, as the reviewer model can be fooled by prompt injection attacks, so it should only be enabled for low-risk workflows with properly configured access lists.

这是一款专门为DeepSeek Harness（DSH）开发的原生社区插件，为DSH添加了基于大语言模型的自动审批功能。对应Codex的approve-for-me功能，它遵循AGENTSCOPE相关设计规范，当开启模型审批模式后，工具调用的权限审批请求会先交给独立的评审模型判断，模型给出允许/拒绝/升级的结论，只有当模型无法决策时，才会将请求转交给人类用户处理，插件在其他审批模式下保持静默。

开启模型审批模式后，插件会先按照确定性路由规则处理审批请求：命中拒绝列表直接拒绝，命中允许列表直接放行，命中人类专属列表则直接转交给人类处理。其他情况则经过熔断器检查，超过配置的连续拒绝阈值后转人类处理，否则由独立评审模型基于工具信息、调用参数和请求原因给出审批结论。适合希望减少低风险工作流中人工审批操作的DSH用户使用。

本插件采用MIT许可证开源，可免费使用和修改。它存在一些已知限制：目前没有专用的机器可读审计事件，暂未实现客户端工具卡徽章UI，仅支持全匹配工具名路由，不支持通配符或参数模式匹配。AI评审不是安全保证，容易被提示注入攻击，仅推荐在低风险工作流中开启，高风险场景仍需保留人工审批环节。

View on GitHub在 GitHub 查看

Recommended强烈推荐 Send to your DSH发给你的 DSH 安装 
请帮我安装这个 DSH 插件。安装前先完成【兼容性检查 + 安全性检查】，检查通过再动手。

插件：dsh-approval-llm（Letter2025/dsh-approval-llm）
仓库：https://github.com/Letter2025/dsh-approval-llm
本站详情页：https://www.yhbd.top/plugins/letter2025-dsh-approval-llm/
本站登记：类型 plugin · 归类 原生 DSH 插件 · 许可证 MIT · ⭐ 7 · 最近提交 2026-09-28 · 主语言 TypeScript

按下面顺序执行，每步先把结论告诉我，再进入下一步：

【1 兼容性检查】
① 我这边：DSH 版本、Node 版本、操作系统、当前 profile（web / desktop）。
② 读它的 README、package.json、插件 manifest，列出它要求的 DSH 版本 / Node 版本 / 操作系统 / 外部依赖 / 需要另外先装的运行时。
③ 逐条比对，结论只写「满足 / 不满足 / 未知」三种；不满足的给出可行替代方案。
④ 检查是否和我已装的插件冲突：命令名重复、skill / tool 重名、端口占用、重复注册的 MCP server。

【2 安全性检查】
① 仓库可信度：和上面「本站登记」是否一致；star / fork 数、创建时间、最近提交，是否归档或长期停更。
② 安装脚本：逐行看 package.json 的 preinstall / install / postinstall，以及 install.sh、setup.ps1 之类脚本。出现 curl|bash、下载后直接执行、混淆代码、访问与插件功能无关的域名，立刻停下来告诉我，不要继续装。
③ 依赖：列出新增依赖，标出无人维护、或与知名包拼写近似的可疑包（typosquatting）。
④ 权限与副作用：它会读写哪些目录、访问哪些域名、需要哪些 DSH 权限（filesystem / network / shell / clipboard 等），以及怎么卸载和回滚。
⑤ 如果它要求 sudo / 管理员权限，或权限明显超出功能所需，先停下来问我。

【3 安装】
上面两步没有「不满足」和「高危项」时才执行；用官方推荐方式安装，不要自行提权。

【4 汇报】
用表格输出：检查项 / 结论 / 依据 / 是否需要我决策。拿不准的一律写「未知」并说明要我怎么确认——不要猜，也不要替我决定。

Send this message to DSH in your current session: it verifies compatibility and security first (answering met / not met / unknown item by item) and only installs once everything checks out — it will stop and ask you if it finds a high-risk item. The box scrolls; the copy is the full prompt. CLI install commands may not be accurate across systems, so DSH is the safer route.把上面这条消息直接发给当前会话里的 DSH：它会先核对兼容性与安全性（逐条给「满足 / 不满足 / 未知」），确认没问题再安装，有高危项会停下来问你。框内可滚动，复制到的是完整提示词；安装命令不一定准确，发给 DSH 更稳。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 1 warning1 项注意
  • Only 7 stars - very few users, little community feedback星标只有 7，几乎没人在用，遇到问题缺少社区反馈
DSH walks through these 9 checksDSH 会逐条核对这 9 项

Compatibility兼容性

  • DSH, Node, OS and profile requirementsDSH 版本 / Node 版本 / 操作系统 / profile 是否满足要求
  • External dependencies and runtimes (Electron / Python / Docker, ...)外部依赖与运行时（Electron / Python / Docker 等）是否齐备
  • Conflicts with installed plugins: command names, skill / tool names, ports, duplicate MCP registration与已装插件是否冲突：命令名、skill / tool 重名、端口占用、重复 MCP 注册

Security安全性

  • Repo matches the facts registered here; archived or abandoned?仓库是否与页面登记一致，是否归档或长期停更
  • Safety of preinstall / install / postinstall and install.sh / setup.ps1preinstall / install / postinstall 与 install.sh、setup.ps1 是否安全
  • curl|bash, download-then-execute, obfuscation, unrelated domains → stop immediatelycurl|bash、下载即执行、混淆代码、无关域名 → 立刻停止
  • Typosquatting or unmaintained packages among the new dependencies新增依赖里有没有 typosquatting 或无人维护的包
  • Requested permissions vs. what the feature actually needs申请了哪些权限、是否超出功能所需（filesystem / network / shell / clipboard）
  • Any sudo / admin requirement, plus uninstall and rollback是否要求 sudo / 管理员权限，以及卸载与回滚方式

Anything uncertain must be marked unknown with a note on how to confirm it. This site's signal screen is a static snapshot, not a security audit.拿不准的必须标「未知」并说明要我怎么确认。本站的信号筛查是静态快照，不能替代安全审计。

Or use CLI install (for developers)或使用命令行安装（适合开发者）

CLI Install命令行安装 

dsh plugin --profile web add dsh-approval-llm

把 Letter2025/dsh-approval-llm 加入你的 DSH 配置（web profile）即可启用。

READMEREADME

15.7 KB · 缓存于 2026-08-26 05:39:59 · 15.7 KB

dsh-approval-llm

release downloads stars license docs Awesome DSH Plugin

Model-based permission approval (approve-for-me) for DeepSeek Harness.

A community plugin that adds a "model approval" permission mode (approve-for-me) to DeepSeek Harness: in that mode, approval/request asks are answered by a separate reviewer model instead of a human — the reviewer decides ALLOW / DENY / ESCALATE, and the request only reaches a human when the reviewer cannot decide or fails. In every other permission mode the plugin stays silent, so human approval is never front-run by the model.

It is the dsh equivalent of Codex's approvals_reviewer=auto_review (--approve-for-me), and it follows the review design of AGENTSCOPE-PLAN-058 / 062 / 063 (orthogonal reviewer dimension, three-way decision, routing policy, prompt isolation, fail-to-human, circuit breaker).

Warning: an AI reviewer is a policy choice, not a security guarantee. It can be fooled by prompt injection from tool output or the agent's own reason. Prefer it for low-risk workflows; keep humanOnlyList, the denyList, and the circuit breaker tight.

Design

Concept (this plugin) Codex AGENTSCOPE-PLAN-058
A dedicated permission mode activates the reviewer approvals_reviewer: auto_review + --approve-for-me UI preset 帮我批准 = DEFAULT + ApprovalReviewer=MODEL
Outside that mode the plugin delegates everything human approval unchanged 请求批准 = DEFAULT + HUMAN
Answer the approval/request waterfall approvals_reviewer: auto_review ApprovalReviewer = MODEL (orthogonal to the permission mode)
Deterministic routing before any model call "deterministic sandbox/network allowlist runs before the guardian" SAFE_ALLOW / DENY / HUMAN_ONLY routing policy
Reviewer model decides ALLOW / DENY / ESCALATE Guardian subagent (Approved / Denied / TimedOut / Abort) ALLOW / DENY / ESCALATE
Reviewer holds an isolated security policy Guardian prompt isolated from the main agent §4.5 prompt isolation
Tool description injected at review time — §4.5 dynamic tool-description injection
Tool arguments recovered from the session log trust layering (arguments reviewed, not just the name) §4.4 argument-level risk
Model failure → hand to human, not counted in the breaker fail-closed guardian §4.8 fail-to-human vs policy decision
Consecutive DENY threshold → hand off to a human circuit breaker (3 consecutive) §4.14 circuit breaker
Explicit provider/model, else the logged conversation route — PLAN-062 per-agent model config + fallback
Decision appended to the session as a user-visible message guardian badge in the UI DENY event pushed to the frontend (§4.17.5)

Showing the opening section of the README — the full document lives in the repository以上为 README 开头摘要，完整文档在仓库内 · View the full README on GitHub →在 GitHub 查看完整 README →

渲染基于仓库 Letter2025/dsh-approval-llm 的 README.md。 在 GitHub 查看原文 →

Related相关推荐

← 上一个 Prev dsh-progressive-tools 下一个 Next dsh-mineru →