@mkaliezz/dsh-agent-doctor
DeepSeek Harness health and safety diagnostics for effective tool surfaces, approvals, and AgentFuse composition
64 results
DeepSeek Harness health and safety diagnostics for effective tool surfaces, approvals, and AgentFuse composition
Installation safety gate & data-protection guard for DeepSeek Harness: 60 static signature rules (31 high/24 medium/5 low) scan plugin sources for malicious install scripts, credential theft, obfuscation, persistence and network callbacks before 'dsh plugin add', plus 12 destructive-command patterns and workspace-boundary checks against accidental deletion. | 安装安全闸门与数据保护闸:60 条静态签名规则(31 高危/24 中危/5 低危)扫描插件恶意行为、持久化与外传,12 种危险命令模式与边界检查拦截误删。
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.
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.
Runtime security guard for DeepSeek Harness (DSH) — blocks command injection, SSRF, credential exfiltration, and destructive operations at runtime.
DeepSeek Harness (dsh) plugin that snapshots your workspace at every turn into a shadow git repository, so you can restore the files to any earlier turn. Your own .git is never touched, and plain directories work too.
Hard-stop DeepSeek Harness tool use after a per-session call budget.
Agent- and GUI-controlled DG-LAB Coyote e-stim plugin for DeepSeek Harness: safety-bounded strength, programmable waveforms, and a DSH-aligned web panel.
Self-protection guardrails for the DeepSeek Harness: protected-path interception, backup-before-destroy, CLI self-recovery commands, and strict sandbox defaulting
Commitment write-gate for AI coding agents: two-tier (deterministic + LLM judge) pre-execution policy. Engine-agnostic core with a DeepSeek Harness (dsh) adapter.
禁止 write 覆盖已存在的文件:write 只允许创建新文件,目标已存在时在工具执行前直接拒绝并提示改用 edit,全局所有会话生效,防止模型用 write 整体覆写已有内容。纯 host 端 dsh 插件。
distro_info: structured wsl -l -v, default vs current distro, UNC path safety.
A dsh policy plugin that gates kubectl writes by kubeconfig context: hard-deny irreversible verbs outside local clusters, ask for the rest.
漂白安全
露营安全应对
徒步安全
用电安全
AI Agent runtime authorization & evidence verification — tool-call GuardrailProvider, CCS 7-dimension verification standard, MCP/DSH security scanner, SSRF/command-injection/credential-exfil blocking with Ed25519 signed receipts.
DSH 回收站插件:最高权限下所有删除操作一律拦截并转入回收站,误删可一键恢复。Intercepts every delete operation (Remove-Item/rm/del/fs.unlink/...) and redirects it to a recoverable trash store.
DeepSeek Harness guardrail that limits how many files an AI coding agent can edit per turn
DeepSeek Harness plugin suite for the heart-health scenario: heart-health preset, resident guidance, controlled heart_* wrapper tools over the heart-algo MCP bridge, and the safety policy that hides the raw mcp__heart-algo__* tools.
Execution-fidelity guard bundle for DeepSeek Harness
DSH 危险命令守卫:通过 tools/pre-execute 门禁拦截 bash/pwsh 高危命令(rm -rf /、mkfs、dd 写设备、管道执行脚本、git push --force 等),支持分级策略(极高危始终拒绝 / 中危跟随全局 deny/ask/log)、用户自定义规则与白名单。
Remote SSH terminal for the DeepSeek Harness web UI: main-area terminal tab, dangerous-command guard (open/guarded/locked), confirmation workflow, and model tools (netshell_servers / netshell_run). Passwords live in the DSH encrypted credential store and never pass through AI sessions.