BlockRunAI/dsh-clawrouter 预览 preview

BlockRunAI/dsh-clawrouter

DeepSeek Harness的安全门:在危险工具调用运行之前,由一个更强的模型进行审查。此外,还支持视觉功能,并且可通过一个钱包使用67个模型,按请求次数通过x402付费。

Project Overview项目介绍

dsh-clawrouter is a native plugin built exclusively for DeepSeek Harness that adds safety review for dangerous commands before they execute. The plugin addresses a common pain point of DeepSeek Harness’ Full Access mode, where users have accidentally deleted entire home directories from unapproved commands. It installs directly via the DSH CLI with the single command dsh plugin --profile web add dsh-clawrouter, and no extra manual setup steps are needed for most standard DSH profiles. When the DSH agent proposes a dangerous command like rm -rf ~, a configured stronger reviewer model checks it and returns allow, deny, or ask for user input, with the result enforced directly by the tool executor.

The plugin also adds a BlockRun provider route that lets users access 75 different models from multiple AI vendors with a single credential. Users can choose between two flexible authentication and payment options: a BlockRun API key from user.blockrun.ai that bills based on exact token usage, or no sign-up at all, paying per request directly via USDC from a Solana or Base wallet using the open x402 micropayment protocol. This plugin is ideal for DSH users who want to run Full Access mode without having to manually approve every command, while still maintaining basic safety guardrails.

dsh-clawrouter is released under the permissive open source MIT license, and has a few documented minor limitations. Multi-step tool use with reasoning models may be slightly degraded because prior-turn reasoning content is not passed back to all supported models. When developing the plugin locally, linking a checked out copy can cause LlmError instanceof checks to fail, so developers are advised to test error codes from a packed npm tarball rather than a linked local checkout.

这是一个专为 DeepSeek Harness (DSH) 开发的原生插件,为 DSH 代理的危险操作增加了执行前审查机制,解决了社区用户对 DSH 全权限模式下误删重要文件的普遍担忧。它会自动调用更强的大模型审查代理准备执行的命令,返回允许/拒绝/询问三种结果,且结果由工具执行器强制执行,而非仅通过提示词生效。可通过 DSH 命令行工具一键安装:dsh plugin --profile web add dsh-clawrouter。

除了命令审查功能,该插件还集成了 BlockRun 模型路由服务,支持通过单一凭证访问 75 种不同厂商的大模型。用户既可以注册获取 BlockRun API 密钥,按实际 token 使用量付费,也可以不注册,直接通过 Solana 或 Base 钱包使用 USDC 按请求付费,付费基于 x402 协议实现。该插件主要面向开启全权限模式、希望兼顾免手动审批和操作安全的 DSH 用户。

该插件采用 MIT 许可证开源,目前已知存在一些小限制:部分推理模型的多步骤工具调用体验可能略有下降,上下文溢出仅通过请求大小检测,不匹配网关错误文本检测器。开发阶段如果使用链接本地检出的方式进行调试,可能会出现 LlmError 类型识别失败的问题,建议使用打包后的 tarball 测试。

Pre-install check安装前体检Compatibility · Security兼容性 · 安全性 1 note1 项提示
  • 21 stars - an early-stage project星标 21,属于早期项目
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-clawrouter

把 BlockRunAI/dsh-clawrouter 加入你的 DSH 配置(web profile)即可启用。

READMEREADME

dsh-clawrouter — review the dangerous command, before it runs

A second brain for your DeepSeek Harness agent

DeepSeek is fast and cheap — keep it for the loop.

This adds what it cannot do: a stronger model reviews the dangerous command before it runs.

75 models from one credential — an API key from user.blockrun.ai, or a Solana or Base wallet if you would rather not sign up.


Review before execute  Claude reviews DeepSeek  API key or wallet  x402 USDC

npm version npm downloads GitHub stars CI TypeScript License: MIT

DeepSeek Harness x402 Protocol Get an API key Solana Base Telegram

English | 中文

dsh-clawrouter is a DeepSeek Harness plugin that puts a stronger model in front of your agent's dangerous actions. When the agent proposes rm -rf ~, a reviewer model reads it and answers allow / deny / ask — enforced by the real tool executor, not by a prompt. It also registers a BlockRun provider route, so the reviewer (and any of 75 models) is reachable from one credential: a BlockRun API key from user.blockrun.ai, billed at exact token usage — or, if you would rather not sign up for anything, a Solana or Base wallet paying per request in USDC over x402. MIT licensed.

dsh plugin --profile web add dsh-clawrouter

Why this exists

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 →

← 上一个 Prev dsh-user-experience 下一个 Next deepseek-harness-desktop →