dsh-tool-vet：装第三方插件之前，先看它会动什么
dsh-tool-vet（@deepwhale-cn/dsh-tool-vet）由 feely0208 维护，面向 admin-security 场景：在安装第三方插件或包之前，对本地目录做静态审查，枚举 install 钩子、凭据路径、混淆执行、出网、外部命令与非 registry 依赖等六类能力，并给出文件行号与规则依据。不做沙箱拦截、不做运行时审批、不替代 dsh-dep-audit。安装：dsh plugin --profile web add @deepwhale-cn/dsh-tool-vet；用法含 --root、--demo、--list-rules、--html。
Read More自创技能安全检测技能：第三方 Skill 八大维度安全扫描
SkillHub 社区技能「自创技能安全检测技能」（@user_c64d3056/sin129，v1.0.0）面向第三方 Skill，从恶意代码、高危调用、硬编码密钥、Prompt 注入、权限越权、依赖安全、数据合规与源码篡改等八个维度扫描，输出风险评分、等级、检测报告与审核建议。无需 API Key。文章说明定位、八维能力、skillhub CLI 安装到 Cursor/Claude Code 等 skills 目录的步骤，以及上线前审核与定期巡检等场景；安装与路径以 SkillHub 官方文档为准。
Read Moredsh-dep-audit：依赖树投毒与供应链风险检查
dsh-dep-audit（@deepwhale-cn/dsh-dep-audit）是 feely0208 维护的 DSH admin-security 插件，扫描当前目录依赖 manifest 与 lock，标出 install/postinstall 脚本、无 lockfile、* / latest、git/http 来源及无上界的 >= 约束等供应链风险面。支持 dsh plugin 安装与 npx 纯 CLI，可 --demo 与 --html 出报告。不查 CVE、不做 typosquat；带 install 脚本的常见包有白名单，命中项需人工确认。
Read Moreskill-scanner：安装前本地静态扫描 Agent Skill 安全风险
skill-scanner 是 SkillHub 社区 Skill（@user_87fe1a85/gwchen17，v1.0.0），由腾讯朱雀 Lab A.I.G（AI-Infra-Guard）支撑，在安装或使用任意 Agent Skill 前做 100% 本地静态安全分析，文件与凭据不外传。无需 API Key，兼容 Cursor、Claude Code、Windsurf、OpenClaw 等。文章说明核心能力、skillhub CLI 安装路径与对话触发语，并区分其与日常开发调试类请求的边界。
Read Moresupply-chain-risk-auditor: Measuring Dependency Supply Chain Risk with Scripts
supply-chain-risk-auditor is an Agent Skill in the Trail of Bits skills marketplace that generates supply-chain risk reports for npm, PyPI, and Go project manifests and lock files. Using collect.py and render.py, it gathers metrics such as version-matched advisories, transitive dependencies, upstream archives, npm publisher concentration, and install scripts, while the Agent handles prioritization and upgrade recommendations. The article explains installation, workflow, the effects of gh authentication, and applicable limitations, emphasizing how to interpret unassessable entries and coverage tables, and cautions against replacing script-based measurements with manual estimates.
Read Moredsh-plugin-guard: DeepSeek Harness Plugin Security Checkup
`dsh-plugin-guard` is a plugin security health-check tool for DeepSeek Harness, designed to identify risks in third-party plugins. It performs tiered plugin assessments through static code scanning, dependency review, AI-powered online auditing, and SafeSkill platform scanning. The tool can detect dangerous APIs, credential leaks, permission declaration mismatches, and supply chain risks, and presents results via a green/yellow/red dashboard. Plugins support version monitoring and alerts, helping developers identify potential security threats before introducing a plugin or during regular security inspections.
Read Moredsh-attestation-proof: Offline Deterministic Release Proof for DeepSeek Harness
dsh-attestation-proof is a plugin for DeepSeek Harness (DSH) used to generate offline, deterministic release attestations. By verifying the consistency of a manifest with local identities and pinned keys, the plugin fills a gap in supply chain security verification. Its core features include offline verification, support for DSSE/in-toto statements, secure report generation, and strict filesystem restrictions. It provides two commands, inspect and verify, and outputs only JSON verdicts containing content without sensitive information, making it suitable for security audits and build quarantine scenarios in air-gapped environments.
Read MoreIntroducing the DeepSeek Harness Plugin: dsh-poison-guard
To prevent supply chain poisoning, DSH Harness (DSH) provides a preinstalled scanner plugin, `dsh-poison-guard`. This tool statically analyzes plugin code and uses a three-layer mechanism to identify threats: AST analysis detects dangerous operations and data leakage; deobfuscation decodes common encodings; and regex-based heuristics match suspicious scripts. After installation, the `scan` command can be used to scan plugin directories, with exit codes supporting CI checks. Note that it performs static analysis only, may produce false positives, relies on Harness’s sandbox configuration to restrict permissions, and cannot defend against heavily obfuscated code.
Read Moredsh-provenance: Provenance verification before DeepSeek Harness plugin installation
dsh-provenance is a DSH plugin developed by Darren Tang, designed to address supply chain security risks during agent plugin installation. By checking the consistency between the pending installation package and the read source code before code execution, it identifies potential issues in advance and provides judgments of "ok," "review," or "block." The tool supports checking source provenance, registry integrity, install hooks, SLSA build provenance, and artifact-to-source comparison. It offers two usage methods: CLI and DSH tool, supports CI integration, and features installation interception to prevent un-audited plugins from being installed. It is worth noting that dsh-provenance focuses on source verification, does not execute audited packages, does not verify Sigstore signatures, and does not perform code behavior scanning or runtime monitoring. Its clean report...
Read MoreEligahyu/dsh-sentinel-scanner: Read-Only Security Scanning for DSH Plugins
Eligahyu/dsh-sentinel-scanner is a read-only static security scanning tool for DeepSeek Harness (DSH) plugins, designed to perform supply chain and security risk assessment before code execution. The tool detects risks such as command execution, credential access, data exfiltration, and obfuscation through 51 AST-based heuristic rules, outputting a risk score from 0 to 100 along with a security verdict, but never executes the scanned code. It offers multiple integration methods: as a DSH Agent Tool plugin, a standalone CLI (`dsh-sentinel`), a GitHub Action, or a JavaScript API. Core features include three scanning modes—source code, package artifacts, and profiles—with support for pre-publishing npm packages.
Read Moredsh-plugin-healthcheck: Install Plugin, Verify First, Then Go Live
dsh-plugin-healthcheck is a DSH plugin health check tool maintained by chenw2759-wq, categorized under admin-security. It provides a "Plugin Inspection" feature in the web settings panel, integrating L0 static checks (file integrity, dependency declarations, high-risk duplicates, cordis usage, etc.), L1 configuration combinations, L2 isolated trial runs (complete boot of child processes without restarting the backend), and C8 pure static trojan scanning. When issues are detected, it supports one-click fixes, automatic rollback (writing home patch to disable lines, HMR hot-reload), or copying predefined prompts for agent use. The ironclad rule for fixes restricts modifications to plugin code and configuration layers only, strictly prohibiting changes to harness source code. Licensed under MIT, with 14 stars on GitHub.
Read More