Kali Security MCP

创建自适应执行上下文

执行自适应策略

获取执行上下文状态

获取自适应执行洞察

智能编排多目标自适应攻击

检查可用的逆向分析工具 - 检测本机逆向工程工具

使用Radare2分析二进制文件 - 开源逆向分析工具

使用Ghidra分析二进制文件 - NSA开源逆向分析工具

自动选择可用工具进行逆向分析 - 智能工具选择

Execute an arbitrary command on the Kali server.

Execute Nuclei vulnerability scanner.

Execute Nuclei CVE vulnerability scan.

Execute Nuclei web application security scan.

Execute Nuclei network security scan.

Execute Nuclei technology detection scan.

Execute DNSrecon for comprehensive DNS enumeration.

Execute WPScan for WordPress security testing.

Execute Reaver for WPS PIN attacks.

Execute Bettercap for network attacks and reconnaissance.

Execute Binwalk for firmware analysis and extraction.

Execute theHarvester for OSINT and information gathering.

Execute Netdiscover for network host discovery.

Execute FFUF web fuzzer (faster alternative to wfuzz).

Execute WhatWeb for web technology identification.

Execute Subfinder for fast subdomain discovery.

Execute HTTP probing for target URLs.

Execute Masscan for ultra-fast port scanning.

Execute Hashcat for GPU-accelerated password cracking.

Search exploit database using searchsploit.

Execute Aircrack-ng for WiFi password cracking.

提交APT攻击链工作流 - 基于知识图谱的智能化并发攻击。

基于目标信息识别攻击面。

生成针对目标的APT攻击路径。

执行APT Web应用攻击链 - 自动化多阶段Web应用渗透。

执行APT网络渗透攻击链 - 自动化多阶段网络渗透测试。

执行APT综合攻击链 - 全面的多向量并发攻击。

启动自适应APT攻击 - 智能化动态调整攻击路径。

获取自适应攻击状态 - 查看攻击进展和发现的信息。

手动触发下一攻击阶段 - 强制进入下一轮攻击。

自适应Web渗透测试 - 智能化Web应用攻击。

自适应网络渗透测试 - 智能化网络攻击。

智能APT攻击活动 - 最高级别的自适应攻击。

Perform authorized attack-surface mapping (non-destructive).

Run authorized network exposure assessment with phase-level output.

Review web exposure through content discovery and service checks.

Run authorized web assessment with exposure review and vuln validation.

Run non-destructive injection verification only (no dump/exfiltration).

Run template-based security validation with constrained severity scope.

Build authorized external asset inventory (subdomains + live hosts).

Run authorized credential audit phase only when allowed_actions permits it.

Run controlled validation phase only when allowed_actions permits it.

Run environment review phase only when allowed_actions permits it.

Run authorized full-chain assessment using neutral external naming.

Start a heavy scan in background; returns job_id immediately.

Poll an async scan job; returns running/done/expired/not_found + result.

Block until an async scan job finishes or timeout_s elapses.

List currently running async scan jobs.

Create an open task on the board.

Claim the best-priority open task matching ``role``.

Complete a task. The result envelope is REQUIRED: summary plus the five supporting keys must all be present (evidence/artifacts/risks/unresolved may be empty lists). Missing keys reject the completion.

Extend a claimed task's lease (only the assignee may renew).

List tasks, optionally filtered by status. Expired leases are swept first (lazy expiry).

Return a compact board overview (agents, open/claimed tasks, stage distribution). Expired leases are swept first.

创建攻击链。

向攻击链添加步骤。

评估攻击链可��性 (0-100分)。

查询攻击链列表。

Run Semgrep static analysis on source code.

Run Bandit security scanner on Python source code.

Run Flawfinder on C/C++ source code to find security vulnerabilities.

Run ShellCheck on shell scripts to find bugs and security issues.

Run comprehensive code security audit using multiple tools.

Execute comprehensive reconnaissance workflow using multiple tools.

启用CTF竞赛模式。

禁用CTF竞赛模式，返回正常渗透测试模式。

创建CTF竞赛会话。

添加CTF题目到当前会话。

获取所有检测到的Flag。

获取所有CTF题目的状态。

CTF快速扫描 - 针对CTF环境优化的快速漏洞发现。

CTF Web攻击链 - 专门针对CTF Web题目的攻击。

Create/open a task workspace and seed target graph nodes.

Get task meta, graph summary, findings counts and recent actions.

Upsert a graph node. next_checks is comma-separated. meta_json is optional JSON object.

Query graph nodes.

List next actions from graph (optionally include insight-sourced checks).

Mark a graph node dead so it stops generating next actions.

Verify a candidate finding by replaying command and matching expected_signal.

Export task report: JSON + Markdown primary; optional docx (python-docx).

Compile handoff/progress for resume (do-not-rescan + next_actions).

Resume plan from last handoff without forcing full rescan.

Observer side-channel: detect duplicate tool calls / empty streaks; advise only by default.

Insight branch: produce graph-bound Hypothesis candidates only (never drives executor).

ATT&CK technique label coverage for task findings (report appendix; not a runtime driver).

Real-time task dashboard: phase, graph size, verified count, recent actions, report paths.

Read-only action_log timeline (what/when). No new write path.

Run surface chain on multiple targets.

LLM 自主安全评估 — 由大模型驱动的 ReAct 循环

Execute ANY tool by name through the command registry (meta fallback).

Execute joomscan for Joomla security testing. Args: target: Target Joomla URL additional_args: Additional joomscan arguments Returns: Joomla scan results