Atlas / MCP servers / ling71671 / open-reverselab

open-reverselabBLOCK

Name: open-reverselab
Author: ling71671

mcp/ling71671/open-reverselab

MCP server by ling71671 · ai agent · ai reverse engineering · android reverse engineering · Audited 2026-09-19

Open-source AI reverse-engineering agent platform and MCP server for Ghidra, Frida, x64dbg and Rizin — automated PE/APK/binary analysis, CTF and malware research, with 100+ MCP tools and a 194-article runnable knowledge base.

Verdict

BLOCK

Grade

Trust score

28 /100

Exposed tools

124 94r · 24w · 6d

Transport

—

License

GPL-3.0

Stars

1,148

Overview

From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.

Open-source AI-powered reverse-engineering platform for autonomous binary analysis, built for Claude Code, Codex, OpenCode and any MCP-compatible agent.

Ghidra · Frida · x64dbg · Rizin · PE · APK · Malware Analysis · CTF · Binary Analysis

An open-source reverse-engineering lab — executable knowledge base, 100+ MCP tools, Agent-native.

From an input signal to an evidence chain, every step is runnable.

[](https://discord.gg/But5j58J2f) [](https://deepwiki.com/LING71671/open-reverselab) [](LICENSE) [](#sponsors)

English · 简体中文

:handshake: Sponsored by Sentry

[Sentry](https://sentry.io) supports openreverselab with a sponsored account — error monitoring and performance tracing for the lab's toolchain.

Since Aug 2026 — see SPONSORS.md for the full sponsor list and how to become one.

What is ReverseLab

ReverseLab is an opinionated, runnable attack-knowledge base for reverse engineers, security researchers, CTF players, and AI Agents. Every article is structured as `Scenario → Input signal → Method → Attack chain → MCP tool mapping`, so a human or an Agent can pick up at any entry signal and walk the chain to evidence.

5 boards spanning web, mobile, Windo

Read from source at commit 83231f9154d0OBSERVED · 2026-09-19

Connect

Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control.

claude-code

claude mcp add reverselab-site -- npx -y [email protected]

claude-desktop

{
  "mcpServers": {
    "reverselab-site": {
      "command": "npx",
      "args": [
        "-y",
        "[email protected]"
      ]
    }
  }
}

Exposed tools (124)

94 read · 24 write · 6 destructive. Blast radius: 6 tools can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.

Tool	Risk	Description
`android_adb_connect`	read	连接 MuMu/Android ADB 端点;默认使用 MuMu 当前实例 serial。
`android_adb_devices`	read	列出当前 adb devices -l。
`android_app_baseline`	read	安装/启动 APK 或已装包,收集 Activity、APK 路径、package info、logcat、Frida 进程基线。
`android_capture_screenshot`	read	抓取当前设备屏幕截图。
`android_clear_logcat`	destructive	清空当前设备 logcat 缓冲区。
`android_crypto_unpack_recipe`	read	Android 解密/去壳 runtime recipe:Frida 抓 Cipher/key/iv/hash/dex loader/dlopen/mmap/RegisterNatives。
`android_current_activity`	read	读取当前前台 Activity。
`android_device_info`	read	读取 Android 设备的 model/sdk/abi/fingerprint/root 状态。
`android_force_stop`	destructive	强制停止 Android app。
`android_frida_ensure_server`	read	在 MuMu/Android root 设备上部署并启动 frida-server。
`android_frida_processes`	read	通过桌面端 Frida 枚举 Android 设备进程。
`android_frida_render_template`	read	渲染 Frida 模板为可直接运行的 JS 源码。
`android_frida_run_script`	write	对指定进程/包运行一次性 Frida JS 脚本,收集 send() 消息并导出 JSON。
`android_frida_status`	read	检查 frida-server 进程和桌面端 Frida 是否能枚举该设备。
`android_frida_template_library`	read	列出可直接复用的 Frida JS 模板。
`android_http_observation_recipe`	read	对已登录/已安装 Android 包做 HTTP/WebView/OkHttp 运行时观察,输出 Frida + logcat 证据汇总。
`android_install_apk`	write	通过 adb install 把 APK 装进 MuMu/Android 设备。
`android_list_packages`	read	列出 Android 包名,可按 query 过滤。
`android_logcat_dump`	read	导出当前设备 logcat 到 exports\\android。
`android_mumu_instance_info`	read	读取 MuMu 当前实例状态、候选 serial 和 adb 可达性。
`android_package_fs_recipe`	read	按包名取证私有目录,列结构并按需回拉 shared_prefs/databases/files 归档。
`android_package_info`	read	导出 dumpsys package 结果,便于分析权限、组件和安装状态。
`android_package_paths`	read	读取包对应的 base.apk / split APK 路径。
`android_pull_artifact_recipe`	read	按包名回拉 APK、截图、包信息、logcat,并写出一份 artifact manifest。
`android_pull_file`	read	从 Android 设备拉取文件到 exports\\android。
`android_pull_package_apk`	read	把设备上的包 APK 拉回 exports\\android\\packages。
`android_push_file`	write	把本地文件推送到 Android 设备。
`android_runtime_file_watch_recipe`	read	对包目录做运行前后快照,输出 shared_prefs/databases/files 的差异清单。
`android_start_package`	write	启动 Android app;未指定 activity 时走 monkey 启动默认入口。
`android_uninstall_package`	destructive	卸载 Android 包。
`burp_launch`	read	生成或可选启动 Burp Suite。默认 launch=False,只返回命令,不弹 GUI。
`burp_status`	read	检查 Burp Suite JAR、wrapper 和默认代理状态;不启动 GUI。
`carve_payloads_from_dump`	read	从 dump/decrypted buffer 中自动 carve PE/DEX payload,并可导入 samples\\unpacked 供下一轮分析。
`copy_sample`	read	复制 samples 下的样本到另一个 samples 位置。
`copy_sample_to_patches`	read	复制样本到 patches 目录,原始样本不变,并写入 audit log。
`ctf-24h-fleet`	read	Multi-target Web CTF 24h loop orchestrator: normalize targets, shard into batches, run ctf-24h-round for each target, synthesize fleet status
`ctf-24h-round`	read	Web CTF 24h loop 的单轮有状态 workflow:初始化/读取 manifest → 执行一轮推进 → 写 checkpoint → 返回 CONTINUE/DONE/EXHAUSTED
`ctf-asset-discovery`	read	CTF digital asset discovery — crt.sh, DNS, WHOIS, HTTP probe, email security, search exposure, synthesis
`ctf-attack-api_business`	read	API and business-logic worker: API discovery, IDOR/BAC, mass assignment, rate limit, payment, signature
`ctf-attack-auth`	read	Auth worker: JWT, OAuth/OIDC, SAML, Host Header, LDAP, session/cookie trust boundaries
`ctf-attack-client`	read	Client-side worker: XSS, CORS, CSP, CSRF, postMessage, WebSocket, Web Crypto, admin bot
`ctf-attack-cve_cloud_dos`	read	CVE/cloud/supply/DoS/database worker: fingerprint-to-CVE, cloud metadata, CI/CD, Kubernetes, DoS, database chains
`ctf-attack-file_ssrf`	write	File and SSRF worker: LFI/path traversal/upload/XXE/file wrappers/SSRF/open redirect
`ctf-attack-injection`	read	Injection worker: SQLi/NoSQLi/SSTI/GraphQL/HPP/CRLF/prototype pollution/gRPC
`ctf-attack-recon`	read	Recon worker: asset discovery, route discovery, JS/API mining, fingerprint, subdomain takeover
`ctf-attack-router`	write	Route Web CTF signals/focus items to concrete attack workflows and run selected workers in parallel
`ctf-dos-assessment`	read	DoS attack surface assessment — map all 13 DoS techniques to discovered targets, execute according to workflow parameters, rate exploitability
`ctf-full-pipeline`	read	CTF 全链路评估流水线 — 资产发现 → DoS攻击面 → 全面漏洞挖掘 → 漏洞逐条验证 → 综合报告
`ctf-vuln-discovery`	read	CTF 全面漏洞挖掘 — 敏感文件/XSS/SQLi/路径穿越/CMS漏洞/认证缺陷/SSRF/WAF绕过,自适应目标技术栈
`ctf-vuln-verify`	read	CTF 漏洞逐条验证 — PoC/漏洞利用/盲注提取/默认口令/CVE确认,基于先前的发现列表
`ctf_autopilot_round`	read	读取 Web CTF ai_manifest.json,执行一轮带 checkpoint 的 autopilot 计划/allowlist 动作。
`ctf_new_challenge`	read	创建新的 CTF 题目 case 目录,从模板初始化。
`ctf_save_request`	write	保存 raw HTTP request 到 exports/ctf-website/<case>/requests,供 Burp Repeater/Intruder 或 sqlmap -r 使用。
`ctf_tool_status`	read	检查 CTF 工具安装状态。
`delete_generated_artifact`	destructive	删除工具生成目录内的文件/目录;默认 dry_run=True。
`delete_sample`	destructive	删除 samples 下样本;默认 dry_run=True。
`die_scan`	read	使用 DiE/diec 扫描文件类型、编译器、packer/signature。默认 JSON 输出。
`extract_frida_buffers`	read	从 Frida JSON 的 data_hex 消息中落盘二进制 buffer,并可自动 carve PE/DEX payload。
`extract_iocs_from_summary`	read	从 Ghidra summary、triage 字符串和分析笔记中提取 IOC 与行为线索,写入 exports\\iocs。
`generate_patch_report`	write	从 mutation audit log 生成 Markdown patch report。
`ghidra_headless_analyze`	read	调用 Ghidra analyzeHeadless 导入并自动分析样本,导出 JSON 摘要;不执行样本。
`ghidra_summary_call_focus`	read	基于 Ghidra xrefs/calls/decompile/imports/strings 给出函数阅读优先级;兼容 legacy summary。
`ghidra_summary_function_detail`	read	按地址或函数名读取单个 Ghidra 函数的 callers/callees/import_refs/string_refs/decompile 证据。
`ghidra_summary_functions`	read	从已导出的 Ghidra summary 中按函数名、signature 或地址过滤 functions。
`ghidra_summary_imports`	read	从已导出的 Ghidra summary 中按 API 名、namespace/library、引用数过滤 imports。
`ghidra_summary_list`	read	列出 exports\\windows\\ghidra 下已导出的 Ghidra summary JSON。
`ghidra_summary_overview`	read	读取 Ghidra summary 的总体信息,不返回完整 functions/imports/strings。
`ghidra_summary_strings`	read	从已导出的 Ghidra summary 中按字符串内容、地址、最小长度过滤 strings。
`hash_file`	read	计算文件 MD5/SHA1/SHA256 和大小。
`http_probe`	read	对目标 URL 发起 GET 探测,收集 header/body/cookie/指纹。
`import_sample`	write	导入文件到 samples,可选 move;适合把外部样本纳入工作台。
`kb_catalog`	read	列出知识库所有板块、分类、条目数和文件数。board 可选: ctf-website\|apk-reverse\|pe-reverse,不传则列出全部板块。
`kb_read_file`	read	读取知识库技术文件内容。自动检测板块或通过 board 参数指定。路径格式如
`kb_router`	read	按攻击信号搜索知识库(支持所有板块),返回匹配的技术文件和路径。board 可选: ctf-website\|apk-reverse\|pe-reverse,不传则全部搜索。
`list_generated_artifacts`	read	列出 exports/patches/projects/reports 中的工具生成物。
`list_samples`	read	列出 samples 下样本,可按子目录过滤。
`make_crypto_replay_scaffold`	read	从 Frida crypto 证据生成可运行的 Python 解密/Hash/HMAC 复现脚本骨架。
`make_pe_crypto_unpack_plan`	read	生成 PE 解密/去壳动态分析包:x64dbg 断点脚本、Windows Frida hook、重点函数队列和 JSON plan。
`make_procmon_filters`	read	根据样本导入表和 Ghidra summary 生成 Procmon 过滤计划,写入 scripts\\procmon。
`make_sigma_stub`	read	根据样本、IOC 结果和可选 summary 生成可继续精修的 Sigma 草案,写入 exports\\sigma。
`make_x64dbg_breakpoint_script`	read	根据 triage/Ghidra summary 生成 x64dbg 断点脚本,写入 scripts\\debug。
`make_yara_stub`	read	根据样本、Ghidra summary 和 IOC 产物生成可继续精修的 YARA 草案,写入 exports\\yara。
`mcp_update_audit`	write	审计工作区/全局 MCP 配置中的本地 Git、npm、PyPI/uvx 工具是否有上游更新或弃用风险。
`move_sample`	write	移动或整理 samples 下的样本。
`mutation_audit_tail`	read	读取 mutation audit log 尾部记录。
`parse_android_crypto_unpack_result`	read	解析 android_crypto_unpack_recipe/Frida JSON,提取 key/iv、crypto op、动态 dex、native loader、mmap/mprotect、RegisterNatives 证据。
`patch_bytes`	write	复制输入文件到 patches 后修改指定 offset 的字节;不原地修改样本。
`patch_pattern`	write	按十六进制 pattern 定位后 patch 副本字节;pattern 支持 ?? 通配。
`patch_pe_bytes`	write	按 PE file offset/RVA/VA 定位后 patch 副本字节;不原地修改样本。
`pe_address_to_offset`	read	将 PE 的 file offset/RVA/VA 映射为文件偏移。
`postprocess_frida_crypto_result`	read	对 Frida crypto/unpack JSON 一键执行 parse、solve、replay scaffold、buffer extract/carve,并输出总 manifest。
`procmon_export_csv`	read	将 exports\\procmon 下的 PML 导出为 CSV,可选加载 Procmon 配置并应用当前过滤器。
`procmon_start_capture`	write	启动 Procmon64 capture,backing file 默认写入 exports\\procmon。
`procmon_stop_capture`	write	停止 Procmon capture。
`project_skills_status`	read	查看项目级 skills/MCP 安装状态和推荐安装命令。
`python_re_tool_install`	write	按 allowlist 安装 Python 逆向库,例如 lief、frida、angr。
`python_re_tool_status`	read	检查 LIEF/Frida/angr/capstone/keystone 等 Python 逆向库是否已安装。
`python_re_tool_version`	read	查询指定 Python 逆向库的版本和模块路径。
`quarantine_sample`	read	把 samples 下样本移入 samples\\_quarantine。
`refine_ioc_sources`	read	将 IOC 条目按 static_confirmed / mixed / note_only 分层,便于后续 YARA / Sigma 精修。
`rename_sample`	write	重命名 samples 下的样本文件。
`rizin_assemble_bytes`	read	调用 rz-asm 把汇编文本转成机器码。
`rizin_assemble_patch`	write	先用 rz-asm 汇编,再对 patches 副本做 patch。
`rizin_bin_info`	read	使用 rz-bin -j -I 输出二进制基础信息。
`rizin_imports`	read	使用 rz-bin -j -i 输出导入表,可限制返回数量。
`rizin_sections`	read	使用 rz-bin -j -S 输出节区信息。
`rizin_strings`	read	使用 rz-bin -zz 输出 raw strings 文本行,可限制返回数量。
`rizin_write_bytes`	write	使用 Rizin raw write 在 patches 副本上写入十六进制字节。
`run_ctf_tool`	write	运行 CTF 工具。tool: sqlmap/dirsearch/jwt_tool/tplmap。args: 命令行参数。
`run_sqlmap_request`	write	用 sqlmap -r 运行已保存的 raw HTTP request。
`sample_autopilot_round`	read	从 battleplan manifest 自动规划/执行下一轮逆向动作:补 Ghidra、生成函数断点、行为断点、unpacking 断点和 Procmon 视图。
`sample_full_workup`	read	一键自动逆向工作流:triage、可选 Ghidra、重点函数队列、x64dbg/Procmon 计划、IOC、YARA/Sigma 草案和总控 manifest。
`search_pattern`	read	按十六进制 pattern 搜索文件,支持 ?? 通配。
`solve_crypto_from_evidence`	read	从 Frida key/IV/input/output evidence 自动尝试常见解密/hash/HMAC,落盘命中结果并可继续 carve PE/DEX。
`toolbox_launch`	read	启动 allowlist 中的 GUI/交互工具,可选打开目标文件;不提供任意 shell。
`toolbox_list`	read	列出 ReverseLab allowlist 工具箱:DiE、Rizin/Cutter、Ghidra、PE-bear、Procmon、x64dbg。
`toolbox_version`	read	对 allowlist 中支持安全版本探测的 CLI 工具执行版本查询。
`triage_pe`	read	组合 hash、DiE、rz-bin info/sections/imports/strings,生成只读初筛结果。
`triage_to_notes`	read	根据 triage 和可选的 Ghidra summary 生成分析笔记骨架,写入 notes\\*.md。
`workspace_copy_artifact`	read	复制 notes/reports/scripts/exports/patches/projects 下的文件或目录。
`workspace_delete_artifact`	destructive	删除 notes/reports/scripts/exports/patches/projects 下的文件或目录;默认 dry_run=True。
`workspace_move_artifact`	write	移动或重命名 notes/reports/scripts/exports/patches/projects 下的文件或目录。
`workspace_read_text`	read	读取 notes/reports/scripts/exports 下的文本文件内容,适合 AI 后续处理。
`workspace_write_text`	write	在 notes/reports/scripts/exports 下创建或更新文本文件。

Trust audit

BLOCKgrade F · trust 28/100 Do not install this without reading the findings. The audit found something that could harm you or your machine.

Layer	What it checks	Result
L0	Provenance & inventory	PASS
L1	Static analysis of the code	FAIL
L2	Instruction surface (what it tells the agent)	FAIL
L3	Class-specific surface	WARN
L4	Behavioural (sandbox)	SKIPPED

What the source does

Filesystem: declared (10 observation(s))
Network: declared (18 observation(s))
Shell: declared (1 observation(s))
Dependencies: not all pinned
Secrets in source: found

Findings (25)

HIGHCode injection · code.eval_exec · CWE-78, CWE-94, CWE-95

.claude/workflows/ctf-vuln-discovery.js:158

'Check for: location.hash/search → innerHTML/document.write, eval() with user input, jQuery .html()/.append() without sanitation, postMessage without origin check.',

Why it matters. evaluates text as code

Fix. remove; use a parser or a dispatch table

HIGHNetwork egress · net.metadata_ip · CWE-200, CWE-319

.claude/workflows/ctf-attack-file_ssrf.js:21

'http://169.254.169.254/',

Why it matters. cloud metadata endpoint: the classic SSRF credential grab

HIGHNetwork egress · net.metadata_ip · CWE-200, CWE-319

.claude/workflows/ctf-vuln-discovery.js:31

'http://169.254.169.254/',

Why it matters. cloud metadata endpoint: the classic SSRF credential grab

HIGHNetwork egress · net.tls_off · CWE-200, CWE-319

scripts/ctf-website/payment_lost_update_probe.py:45

p.add_argument("--insecure", action="store_true", help="关闭 TLS 证书校验")

Why it matters. certificate verification is disabled

Fix. leave verification on

HIGHPrompt injection · prompt.override · CWE-94, CWE-1427

kb/general/techniques/attack-network.md:95

JAILBREAK["Jailbreak<br/>ai-security"]

Why it matters. asks the agent to drop prior instructions or safety