Security scanner for DSH plugins: static and semantic passes over plugin source, DSH-specific attack-surface rules, npm audit, batch scanning, and an HTML report with per-finding severity and evidence.
The SEO audit toolkit built for DeepSeek Harness (DSH) — runs locally, no Claude Code, no API key required. Self-contained port of claude-seo v2.2.5: 53 Python scripts + 24 sub-skills + 18 agents, weighted score, gated fan-out, schema.org, E-E-A-T, GBP, GEO/AI Overviews.
Claude Code-style auto approval layer for DeepSeek Harness: deterministic rules + two-stage classifier, circuit breaker, fail-to-human. Shadow mode day one.
PSA is a multi-classifier behavioral analysis engine for LLM responses. It classifies every sentence into behavioral postures, then derives metrics from the posture sequences to detect adversarial stress, sycophancy, hallucination risk, persuasion techniques, and input pressure — in real time.
模式匹配自动注入 steering rules,不占系统上下文 - Inject rules when needed, without wasting context. Similar to oh-my-pi's "Time-traveling stream rules", but with a very simple and compact code implementation.
DeepSeek Harness 自动审批插件:在 workspace-write 沙箱之上叠加确定性规则 + LLM 安全审批,自动模式不放宽沙箱、fail-closed。 Safe auto-approval for DeepSeek Harness — deterministic rules + LLM review on top of the workspace-write sandbox. Auto mode without ever granting full-access.