插件目录 / Security / dsh-security-suite
dsh-security-suite
未验证 Zenquiem
功能简介
DeepSeek Harness安全评估工作流
未验证 — 尚未实测
DeepSeek Harness安全评估工作流 尚未验证——请自行安装测试。
「未验证」表示我们的自动化 CI 尚未安装过该插件。功能描述与版本兼容性均为作者声明。这不是安全审计,也不代表对第三方代码的背书。
README
DSH Security Suite
A native security assessment suite for
DeepSeek Harness (DSH),
adapted from the architecture of
openai/codex-security (Apache-2.0).
It runs repository, diff, and deep security scans inside DSH: LLM review
subagents do the discovery (baseline auditors, focused investigators,
file-review and deep workers), while a deterministic engine produces the
evidence baseline, and every reportable finding is bound to a claim-token
validation receipt and an attack-path receipt before it can be finalized.
Quick start
npm install
npm run build
dsh plugin --profile web add /absolute/path/to/dsh-security-suite
dsh run "Run a deep security scan, validate candidates, trace attack paths, and produce a report for this workspace."
Capabilities
| Workflow | What it does |
|---|---|
| Standard scan | One independent baseline auditor plus focused investigator DSH subagents over source-backed investigation packets (security_scan, default discovery: llm) |
| Diff scan | One restricted file-review subagent per changed file, anchored to the changed code (security_review_diff, default discovery: llm) |
| Deep scan | Six workers per round, each with a distinct review lens, a semantic reducer per round, and stop-after-no-new saturation |
| Deterministic fallback | The rule/AST/flow engine remains available via discovery: engine and still produces the receipt baseline |
| Validation & attack paths | Claim-token-bound validation and attack-path receipts; isolated runtime evidence with snapshot checks |
| Tracking | GitHub / Jira / Linear issues and private GitHub draft advisories with preview, duplicate lookup, and readback |
| Triage & backlog | Import GitHub REST findings or Jira/Linear tickets, triage against local source evidence |
| Remediation | Reviewed, approval-gated, atomic multi-file patches with rollback and verification scans |
| Exports | Markdown, JSON, SARIF, CSV; canonical scan-manifest.json / findings.json / coverage.json |
| Extras | Threat models, vulnerability write-ups, hardening portfolios, disclosure campaigns, bulk scans, pre-commit hook |
See the full tool surface and evidence model in
docs/evidence-contract.md.
How discovery works
- Standard (
discovery: llm, default): the plugin freezes an in-scope
worklist from scan receipts, launches one independent baseline auditor
subagent, builds a source-backed threat map, groups review questions into
investigation packets, and runs one focused investigator subagent per
packet. Every candidate must cite a receipted in-scope location. - Diff (
discovery: llm, default): each changed source file is assigned
to one restricted file-review subagent that reads it in full. - Deep: six workers per round with distinct lenses (forward dataflow,
backward from sinks, authorization logic, open-ended, parsers, secrets)
review immutable 200-line regions; a semantic reducer subagent merges
equivalent candidates across workers; saturation requiresstopAfterNoNewconsecutive zero-novelty rounds. - Engine (
discovery: engine): the deterministic rule/AST/flow engine for
JS/TS, Python, Go, Java, C#, PHP, Ruby, C, C++, and Rust.
Discovery never auto-confirms vulnerabilities. A finding becomes reportable
only after a source-backed validation receipt and a separately claimed
attack-path receipt.
Configuration
config:
enabled: true
maxFiles: 500
maxFileBytes: 262144
deepScan: # deep-scan engine (codex-security semantics)
workers: auto # positive integer, or 'auto' (capped at 6)
stopAfterNoNew: 6 # saturate after this many zero-novelty rounds
maxDiscoveryRuns: 60
maxTimeHours: 96
knowledgeBase: [] # files/dirs of Markdown or plain text, read-only
scanPrompt: '' # extra instructions for every LLM discovery worker
State is placed under DSH_SECURITY_SUITE_STATE_DIR or stateDir
(default ~/.dsh-security-suite); the engine confines repository and
knowledge-base paths to the active workspace.
Documentation
- docs/evidence-contract.md — the evidence model,
receipts, coverage semantics, and per-rule analysis boundaries - MIGRATION.md — the codex-security capability matrix and
migration status
Attribution
Worker prompt templates, canonical scan-contract field trees, severity policy,
and deep-scan configuration semantics are adapted from
openai/codex-security (Apache-2.0).
No third-party security-runtime source code or dependency is executed.
License
Apache-2.0.
安装
装一次目录插件,之后本站所有插件都能让 DeepSeek Harness 自动找、自动装:
dsh plugin add dshbase-catalog 然后对 agent 说「帮我装 dsh-security-suite」,它会在目录里找到并自动安装。文档:dshbase-catalog · 已验证场景包。
该插件是 GitHub 源码(未发 npm)——直接从仓库装:
Web profile:
dsh plugin --profile web add github:Zenquiem/dsh-security-suite Headless(CLI)profile:
dsh plugin --profile headless add github:Zenquiem/dsh-security-suite 实测报告
尚未 L3 验证——若已跑过,见下方失败备注。
备注:验证: runtime-fail (0.1.0-rc.6) 浏览全部待验证失败 →
使用场景
加固 agent 或其工作区——扫描、净化或审计——先拦下不可信内容和代码。
适合谁
在意供应链和提示注入风险、想要内置防护的人。
二次开发建议
检测器和策略是缝——加规则、作用域或更丰富的被标记项审计日志。