safety 15

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

Harness-neutral reliability infrastructure for AI coding agents: reversible destructive actions, pre-emission redaction, recovery evidence, and a shared Decision Protocol with optional native adapters.

Self-protection guardrails for the DeepSeek Harness: protected-path interception, backup-before-destroy, CLI self-recovery commands, and strict sandbox defaulting

DeepSeek Harness bundle that refuses shell commands terminating browsers or the harness's own host processes by image name.

DSH plugin for smart checkpoints, workspace safety guards, and instant rollback

A dsh policy plugin that gates kubectl writes by kubeconfig context: hard-deny irreversible verbs outside local clusters, ask for the rest.

Independent two-layer safety gate around the DeepSeek Harness agent loop: deterministic and model-classifier verdicts for prompts, streamed output, tools, and tool results

Codex-style auto-approval for DeepSeek Harness. An independent reviewer decides allow/deny, fail-closed, with a rationale for every decision. Engine: an LLM route, or TypeSafe's Jev (`reviewer.engine`). · DSH 插件：独立复核器裁决，引擎可选 LLM 路由或 Jev。

Host-only command safety gate for DeepSeek Harness: blocks recognized dangerous operations and requests DSH approval when a command cannot be inspected.

安全技术交底记录要素齐备性与签字闭环核对（依据公开的安全生产法规与检查标准，仅提示差异，不作出定性结论）

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

Consequence analysis for DeepSeek Harness: it says what an extra-permission call costs you — the device, your delegated authority, your wider interests. State-based, not guessed.

Pre-execution safety valve for DeepSeek Harness: a TypeSafe Jev question decides whether a shell command/script would irreversibly delete or overwrite real data. Four states (allow/revise/block/escalate), mounted on tools/pre-execute. Zero dependencies; W

Auto-reject unanswered permission requests with a model-visible timeout notice

Gear-based autonomy governor for the DeepSeek Harness agent: the entropy-sdk EntropyRuntime core (gear ladder G0-G4, utility gate, slow-up/fast-down policy, tamper-evident audit chain) bound to the Harness tool registry, with a live Web UI and paper-ready