dsh-purge
DeepSeek Harness 破甲:让所有模型都能破甲,不同模型可换不同提示词;默认提示词面向国模「小码酱」。Jailbreak for every model — swap prompts per model. 求 Star 收藏 ⭐ 如要二开,请查看署名要求部分
21 results
DeepSeek Harness 破甲:让所有模型都能破甲,不同模型可换不同提示词;默认提示词面向国模「小码酱」。Jailbreak for every model — swap prompts per model. 求 Star 收藏 ⭐ 如要二开,请查看署名要求部分
Read-only security & compliance toolkit for DeepSeek Harness: prompt-injection detection (rule engine with a pluggable model classifier), Chinese-PII and structured-JSON redaction, and a local configuration security audit that emits redacted, reproducible, self-checksummed risk reports.
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.
CC-style auto mode for DeepSeek Harness: deterministic deny/allow rules + pre-execute gate + model-agnostic two-stage classifier. Two-state (allow/reject) classifier since 0.8.0. TypeScript rewrite merging dsh-auto-mode v0.4.1 with Nuo-cl/dsh-auto-mode native integration.
DSH sandbox extension: keep declared workspace subpaths (e.g. .git) read-only, honor a read-only rules file at the workspace root, grant extra writable roots under workspace-write, and let the model request session-scoped write access; enforced for sandboxed CLI commands and the write/edit tools. 为 DSH 沙箱增加工作区子路径只读保护 (例如 .git), 支持工作区根下的只读规则文件, 可在 workspace-write 下声明额外可写根, 并允许模型申请本会话的可写授权; 命令沙箱与 write/edit 工具都生效.
Tiered auto-review for DeepSeek Harness (DSH): static-rule safety net + LLM reviewer + human fallback. Auto-approve safe actions, auto-deny irreversible ones, ask a human for the rest. PURE VIBE CODING - not audited, use at your own risk.
Strict check: verify code and commands against real checkers instead of reading them — Lean 4 kernel checking with an axiom audit, language type/syntax checks, and static defect rules.
Maestro Guard — pre-execute safety gate: native DSH approval prompts, parsed command-surface rules, decision journal
A DeepSeek Harness (dsh) bundle that turns code review into a hard gate: deterministic severity rules, LLM-assisted findings, team approval quorum and a compliance audit trail.
Portable DSH WebUI skill security inspection with local rules and optional model review
DSH Cordis plugin: turn arbitrary source text or images with text, diagrams, and tables into an AI-generated knowledge graph (facts / inferences / concepts / definitions / examples / counter-examples / rules) with two-way linking between the graph and the original text, plus evidence-grounded verification, questioning, one-click fixes and an audit log.
DSH Auto Guard: a Claude Code Auto Mode-like command approval mechanism that adds an LLM safety net on top of full access, with rules, caches, file tracker, and sensitive path gates.
Agent harness 配置系统:agents/skills/hooks/rules 四层分工、instincts 持续学习(自动学模式、聚簇成技能)、AgentShield 安全审计(secrets/权限/hook 注入/MCP 风险)、Unified Memory Vault 跨 harness 共享、GateGuard 破坏性命令门卫。受 affaan-m/ECC(185k★ MIT)启发。
DSH 规则执行引擎 v3:容器解析 AGENTS.md + 理解器 + 匹配机 + 执行框架
Static security audit for dynamic Cordis plugins in DeepSeek Harness (DSH): rule-based source scanning, risk reports injected into tool results, and user-approval escalation before activating high-risk plugin packages.
Four sequential DNS gates (format, dead domain, MX, role/placeholder) classify email addresses as sendable, risky, or dead — with a hard integrity rule: every verdict traces to a query run in that session.
Windows notification-area notifications for DeepSeek Harness: permission requests, branch choices, and task completion, plus a standing one-item-at-a-time todo status rule.
Installation safety gate & data-protection guard for DeepSeek Harness: 60 static signature rules scan plugin sources for malicious install scripts, credential theft, obfuscation and network callbacks before 'dsh plugin add', with scan baselines (gate_diff), structured reports (gate_report), 12 destructive-command patterns and workspace-boundary checks.
Unofficial DeepSeek Harness plugin that triages sandbox-escalation approvals: hard rules plus a second-model review. Stage 2: observe-only; every prompt still reaches you.
DSH permission-gate for the DeepSeek Harness 0.1.5 line: a single self-sufficient, deterministic-first, fail-closed gate covering P0 hard-deny -> P1 session grant -> P2 static rule (allow/deny) chain -> P3 optional LLM semantic classifier -> P4 ask, with
Declarative Claude Code-style permission rules plus a Codex-style process-level network policy for DeepSeek Harness: ordered allow/deny/ask rules with tool-name, argument (glob/regex), workspace-path, and network-target (domain/ip/port/scheme) matching on