Qu1cksc0pe

All-in-One malware analysis tool for analyze many file types, from Windows binaries to E-Mail files.

You can get:

• What DLL files are used.
• Functions and APIs.
• Sections and segments.
• URLs, IP addresses and emails.
• Android permissions (Dangerous/Special/Info).
• MITRE ATT&CK mappings (Windows + Linux static analysis).
• File extensions and their names.
• Embedded executables/exploits.

And so on...

Qu1cksc0pe aims to get even more information about suspicious files and helps user realize what that file is capable of.

Qu1cksc0pe Can Analyze Currently

Files	Analysis Type
Windows Executables (.exe, .dll, .msi, .bin)	Static, Dynamic
Linux Executables (.elf, .bin)	Static, Dynamic
MacOS Executables (mach-o)	Static
Android Files (.apk, .jar, .dex)	Static, Dynamic(for now .apk only)
Golang Binaries (Linux)	Static
Document Files	Static
VBScript/VBA Family (.vbs, .vbe, .vba, .vb, .bas, .cls, .frm)	Static (--docs)
HTML Documents (.html, .htm)	Static (--analyze)
JavaScript (.js)	Static (--analyze)
HTA / HTML Application (.hta)	Static (--analyze)
Windows Batch Scripts (.bat, .cmd)	Static (--analyze)
Windows Shortcut (.lnk)	Static (--analyze)
Archive Files (.zip, .rar, .ace)	Static
PCAP Files (.pcap)	Static
Powershell Scripts	Static
E-Mail Files (.eml)	Static

MCP Server

Qu1cksc0pe ships an MCP server (Modules/mcp_server.py) that exposes its static-analysis features as tools for MCP-aware clients (Claude Code, Claude Desktop, etc.). It shells out to qu1cksc0pe.py the same way the Web UI does, so it needs no code changes to stay in sync with the CLI, and it only imports the mcp package itself at startup (the individual analyzers' own dependencies are only needed once a tool actually runs).

Install the extra dependency (already included in requirements.txt):

pip install "mcp>=2.0.0"

Launch it through the --mcp flag, same as every other Qu1cksc0pe command:

python3 qu1cksc0pe.py --mcp

Transport defaults to streamable-http (binds 127.0.0.1:8765/mcp), so the server is a persistent process any number of clients can attach to and detach from independently -- run it once in its own terminal, point clients at http://127.0.0.1:8765/mcp. Override with:

SC0PE_MCP_TRANSPORT=stdio python3 qu1cksc0pe.py --mcp   # traditional one-client-per-process model

Env var	Default
SC0PE_MCP_TRANSPORT	streamable-http	streamable-http, stdio, or sse.
SC0PE_MCP_HOST	127.0.0.1	Bind address for streamable-http/sse.
SC0PE_MCP_PORT	8765	Bind port for streamable-http/sse.
SC0PE_MCP_HTTP_PATH	/mcp	URL path for streamable-http.

A project-level .mcp.json is included so Claude Code picks the server up automatically for this repo. It pins stdio explicitly (via env), since Claude Code spawns and owns a fresh process per session rather than attaching to one you started yourself:

{
  "mcpServers": {
    "qu1cksc0pe": {
      "command": "python3",
      "args": ["qu1cksc0pe.py", "--mcp"],
      "env": { "SC0PE_MCP_TRANSPORT": "stdio" }
    }
  }
}

If python3 on your PATH isn't the interpreter with Qu1cksc0pe's dependencies installed (common on Windows, or with multiple Python installs), change command to the full path of the right python/python.exe, or run python3 -c "import mcp" first to check.

Tools: analyze_file, analyze_document, analyze_archive, detect_packer, detect_language, extract_iocs, check_resources, check_signatures, scan_hash, scan_virustotal, configure_virustotal_api_key, configure_ai_api_key, update_hash_database, list_supported_file_types. Each tool validates its input file/folder locally (rejecting files >= 50MB, since the CLI would otherwise prompt interactively) before invoking the CLI, and returns the resulting JSON report(s) plus captured console output. Interactive-only features (--watch dynamic analysis, --ui, --install) are intentionally not exposed as tools.

The five analysis tools that support ai=True also take an ai_provider argument ("auto"/"ollama" (default, local), "claude", "openai", "deepseek", "kimi", or "glm") -- see AI Analysis Providers below. Configure a cloud key first with configure_ai_api_key(provider="claude", api_key="...") (or whichever provider).

Logs: every tool call and CLI dispatch (command, duration, exit code, reports collected) is logged to stderr and to sc0pe_reports/mcp/mcp_server.log. Set SC0PE_MCP_LOG_LEVEL=DEBUG for full stderr output too, or SC0PE_MCP_LOG_FILE=0 to disable the file sink.

AI Analysis Providers

--ai (and the MCP tools' ai=True) summarizes a generated report with an LLM. Ollama (local) is the default; five cloud backends are also supported.

Provider	Flag/value	Env var
Ollama (default)	auto or ollama	OLLAMA_HOST
Claude (Anthropic)	claude	ANTHROPIC_API_KEY
OpenAI	openai	OPENAI_API_KEY
DeepSeek	deepseek	DEEPSEEK_API_KEY
Kimi (Moonshot AI)	kimi	MOONSHOT_API_KEY
GLM (Zhipu AI)	glm	ZHIPUAI_API_KEY

Ollama needs no key -- install Ollama and select the model via [Ollama] model in Systems/Multiple/multiple.conf. For a cloud provider, either set its env var above, or save a key through the interactive key manager:

python qu1cksc0pe.py --key_init
# >>> Qu1cksc0pe API Key Manager
#   1) VirusTotal
#   2) Claude (Anthropic)
#   3) OpenAI
#   4) DeepSeek
#   5) Kimi (Moonshot AI)
#   6) GLM (Zhipu AI)
#   0) Exit

--key_init --key_provider <name> (e.g. --key_provider claude) skips the menu and prompts for just that one key -- useful for scripts (this is what the MCP server's configure_ai_api_key/configure_virustotal_api_key tools do under the hood).

# Explicit provider selection (auto/ollama is the default -- no flag needed for local analysis)
python qu1cksc0pe.py --file suspicious_file --analyze --ai --ai_provider claude

The default (auto/unset) is unchanged from prior versions: Ollama, falling back to a heuristic summary if it's unavailable. Cloud providers are strictly opt-in -- report data is only sent off-machine if you explicitly pass --ai_provider <name> or set SC0PE_AI_PROVIDER. See the Environment Variables table for model/base-URL/timeout/token tuning per provider.

Usage

python qu1cksc0pe.py --file suspicious_file --analyze

# Launch Web UI
python3 qu1cksc0pe.py --ui

Screenshot

Updates

12/08/2026

• New feature: added an MCP server (--mcp, Modules/mcp_server.py, .mcp.json) exposing Qu1cksc0pe as tools for MCP clients like Claude Code, defaulting to streamable-http (persistent, multi-client; stdio/sse also available) with logging to stderr and sc0pe_reports/mcp/mcp_server.log (SC0PE_MCP_LOG_LEVEL/SC0PE_MCP_LOG_FILE). See the "MCP Server" section above.
• New feature: --ai now supports Claude, OpenAI, DeepSeek, Kimi, and GLM as alternative backends to Ollama via --ai_provider/SC0PE_AI_PROVIDER (or the MCP tools' ai_provider argument). Ollama stays the local-only default unless a cloud provider is explicitly selected; see "AI Analysis Providers" above.
• New feature: --key_init is now an interactive menu covering VirusTotal + all five AI providers instead of a single VirusTotal-only prompt. --key_init --key_provider <name> skips the menu for scripted use.
• Bug fix: --key_init saved (and could silently overwrite) an empty API key if you pressed Enter without typing anything. Empty input is now rejected.
• Bug fix: execute_module()'s os.system() call mis-quoted its command on Windows, silently breaking every analysis whenever the interpreter or project path contained a space.

Available On

Recommended Systems

• Parrot OS
• Kali Linux
• Windows 10 or 11

And also another Linux distributions like as Kali/Parrot

Setup and Installation

[!NOTE]
If you encounter issues with the Python modules, creating a Python virtual environment (python_venv) should resolve them.
For detailed setup and troubleshooting (dependencies, Docker usage, Windows notes), see the project overview documentation.
AI model selection is manual: set [Ollama] model in Systems/Multiple/multiple.conf to the exact model you want to use.

# First you need to clone Qu1cksc0pe with this command
git clone --depth 1 https://github.com/CYB3RMX/Qu1cksc0pe

# After cloning the repository YOU MUST create a python virtual environment (for handling python modules)
virtualenv -p python3 sc0pe_venv
source sc0pe_venv/bin/activate

# You can simply execute the following command it will do everything for you!
bash setup.sh
#
# setup.sh also installs required system tools (e.g. adb, strings, unzip, 7z) and sets up JADX.

# If you want to install Qu1cksc0pe on your system just execute the following commands.
python qu1cksc0pe.py --install # Optional

# To prevent interpreter errors after installation, use dos2unix.
dos2unix /usr/bin/qu1cksc0pe

# Or you can use Qu1cksc0pe from Docker!
docker build -t qu1cksc0pe .
docker run -it --rm -v $(pwd):/data qu1cksc0pe:latest --file /data/suspicious_file --analyze

# For Windows systems you need to execute the following command (Powershell)
# PS C:\Users\user\Desktop\Qu1cksc0pe> .\setup.ps1
#
# setup.ps1 handles winget dependency fallback, Python + 7-Zip setup,
# Sysinternals strings EULA acceptance, and resilient Ollama installation.
# If cloud model auth is needed, run:
# ollama signin
# ollama pull kimi-k2.5:cloud

Environment Variables

You can change some analyzer behaviors via environment variables (useful for CI, reproducibility, or controlling report size/timeouts).

Linux/macOS (bash/zsh) example

SC0PE_ANDROID_REPORT_DETAILED=1 python qu1cksc0pe.py --file app.apk --analyze --report

Windows (PowerShell) example

$env:SC0PE_ANDROID_REPORT_DETAILED="1"
python .\\qu1cksc0pe.py --file app.apk --analyze --report

Variable	Default	What It Does
SC0PE_ANDROID_REPORT_DETAILED	0	Android analyzer JSON becomes more verbose (keeps larger fields and higher limits). Includes more details under resource_scan, and keeps large duplicate fields like code_patterns more often.
SC0PE_WINDOWS_REPORT_DETAILED	0	Windows analyzer stores per-category API lists in more detail (instead of unique API names only).
SC0PE_AUTO_DECRYPT_CHAIN	0	Document analyzer: when an Office document decryption succeeds, automatically re-runs analysis on the decrypted output (best-effort).
SC0PE_AI_INTERESTING_PATTERNS_MAX_KEYS	25	AI analyzer: limit how many keys from interesting_string_patterns are included in the LLM prompt.
SC0PE_AI_INTERESTING_PATTERNS_MAX_VALUES	30	AI analyzer: limit list size per interesting_string_patterns key in the LLM prompt.
SC0PE_AI_INCLUDE_TEMP_EXCERPT	0	AI analyzer: include raw temp.txt excerpt in prompt when set to 1 (default is parsed/summarized mode without raw excerpt).
SC0PE_AI_TEMP_TXT_EXCERPT_CHARS	800	AI analyzer: character limit for raw temp.txt excerpt (used when SC0PE_AI_INCLUDE_TEMP_EXCERPT=1).
SC0PE_AI_TEMP_TXT_MAX_STRINGS	50	AI analyzer: limit number of meaningful strings selected from parsed temp.txt.
SC0PE_AI_TEMP_TXT_MIN_LEN	6	AI analyzer: minimum length for a meaningful string extracted from temp.txt.
SC0PE_AI_TEMP_TXT_MAX_LEN	180	AI analyzer: maximum length for a meaningful string extracted from temp.txt.
SC0PE_AI_TEMP_PARSE_MAX_BYTES	2097152	AI analyzer: max bytes to parse from temp.txt while building compact evidence.
SC0PE_AI_TEMP_PARSE_MAX_LINES	12000	AI analyzer: max lines to parse from temp.txt.
SC0PE_AI_TEMP_SAMPLE_LINES	2500	AI analyzer: sample size used for meaningful-string scoring.
SC0PE_AI_TEMP_IOC_CAP	40	AI analyzer: cap for IoC candidates parsed from temp.txt.
SC0PE_AI_TEMP_IOC_PROMPT_MAX	20	AI analyzer: max parsed IoC values per kind sent to LLM prompt.
SC0PE_AI_MAX_REPORT_CHARS	180000	AI analyzer: threshold for full-report prompt mode; larger reports are compacted automatically.
SC0PE_AI_COMPACT_MAX_LIST_ITEMS	40	AI analyzer: list sampling limit in compact report mode.
SC0PE_AI_COMPACT_MAX_STR	220	AI analyzer: max string length per field in compact report mode.
SC0PE_AI_COMPACT_MAX_DEPTH	4	AI analyzer: nested depth limit in compact report mode.
SC0PE_AI_OLLAMA_HTTP_TIMEOUT	60	AI analyzer: Ollama HTTP call timeout (seconds).
SC0PE_AI_HTTP_PROBE_TIMEOUT	20	AI analyzer: short probe timeout before full HTTP generation call (seconds).
SC0PE_AI_OLLAMA_CLI_TIMEOUT	90	AI analyzer: Ollama CLI call timeout (seconds).
SC0PE_AI_TOTAL_BUDGET	120	AI analyzer: total generation budget across retries/fallback attempts (seconds).
SC0PE_AI_OLLAMA_NUM_PREDICT	700	AI analyzer: default generation token budget per Ollama call.
SC0PE_AI_OLLAMA_RETRY_NUM_PREDICT	1400	AI analyzer: generation token budget for retry when output looks truncated.
SC0PE_AI_OLLAMA_NUM_CTX	8192	AI analyzer: Ollama context window setting.
SC0PE_AI_DISABLE_THINK	1	AI analyzer: sends think=false (if supported) and removes thinking artifacts from displayed/saved output.
SC0PE_AI_ALLOW_MODEL_FALLBACK	1	AI analyzer: when 1, can try locally available Ollama models if configured model fails/unavailable.
SC0PE_AI_SKIP_CLOUD_WHEN_LOCAL	1	AI analyzer: prefer local models over cloud-tagged models when local options exist.
SC0PE_AI_SKIP_CLOUD_CLI	1	AI analyzer: skip cloud-tagged models for CLI fallback attempts.
SC0PE_AI_MAX_MODEL_CANDIDATES	4	AI analyzer: maximum number of candidate models to try in fallback chain.
SC0PE_AI_PROVIDER	auto	AI analyzer backend: auto/ollama (default, local, nothing below applies), claude, openai, deepseek, kimi, or glm. Same as --ai_provider; the flag wins if both are set.
SC0PE_AI_CLAUDE_MODEL	claude-haiku-4-5-20251001	AI analyzer: model used when SC0PE_AI_PROVIDER=claude.
SC0PE_AI_<PROVIDER>_MODEL	see below	AI analyzer: model for openai/deepseek/kimi/glm (e.g. SC0PE_AI_DEEPSEEK_MODEL). Defaults: openai=gpt-4o-mini, deepseek=deepseek-chat, kimi=moonshot-v1-8k, glm=glm-4-flash. Cloud model names drift; override if stale.
SC0PE_AI_<PROVIDER>_BASE_URL	provider default	AI analyzer: chat-completions endpoint for openai/deepseek/kimi/glm, in case a provider changes its API URL.
SC0PE_AI_CLAUDE_MAX_TOKENS / SC0PE_AI_<PROVIDER>_MAX_TOKENS	1200	AI analyzer: response token cap for the respective cloud provider.
SC0PE_AI_CLOUD_HTTP_TIMEOUT	90	AI analyzer: HTTP timeout (seconds) for any cloud provider call.
ANTHROPIC_API_KEY / OPENAI_API_KEY / DEEPSEEK_API_KEY / MOONSHOT_API_KEY / ZHIPUAI_API_KEY	unset	AI analyzer: cloud provider API keys. Take precedence over keys saved via --key_init.
SC0PE_AI_FILTER_WHITELIST_DOMAINS	1	AI IoC sanitizer: filter legit/whitelisted domains using Systems/Multiple/whitelist_domains.txt.
SC0PE_AI_ALLOW_SHORT_DOMAINS	0	IoC sanitizer: allow very short SLD domains (disabled by default to reduce false positives).
SC0PE_AI_MIN_SLD_LEN	4	IoC sanitizer: minimum registrable-label length for domain validation.
SC0PE_AI_ALLOW_FILELIKE_TLDS	0	IoC sanitizer: when 0, filters file-like pseudo-domains such as sheet1.xml.
SC0PE_AI_KEEP_LOCAL_PATHS	0	IoC sanitizer: when 0, removes local analysis machine paths from file_paths.
SC0PE_DOC_AUTO_EXTRACT_MACROS	1	Document analyzer: automatically extract detected VBA/XLM macros into report output (0 disables).
SC0PE_REPORT_MAX_MACRO_CHARS	50000	Document analyzer: per-macro text cap used while saving extracted macro content into JSON report.
SC0PE_EMAIL_DNSBL_FILTER_NOISY	1	Email analyzer: filter noisy DNSBL providers to reduce false positives.
SC0PE_EMAIL_DNSBL_ALLOW_UNKNOWN	0	Email analyzer: include/exclude DNSBL hits with unknown category.
SC0PE_EMAIL_DNSBL_NOISY_PROVIDERS	unset	Email analyzer: comma-separated extra DNSBL providers to treat as noisy.
SC0PE_AUTO_CLEANUP_ATTACHMENTS	unset	Email analyzer: set 1 for auto-delete, 0 for never-delete, unset for interactive prompt.
OLLAMA_HOST	http://127.0.0.1:11434	AI report analysis backend (Ollama). Set this if Ollama is remote or on a different host/port.
JAVA_HOME	unset	Android analyzer: helps locate Java runtime for JADX. Set this if Java is installed but not detected.

Static Analysis

Normal analysis

Description: You can perform basic analysis and triage against your samples.

Usage: python qu1cksc0pe.py --file suspicious_file --analyze

Resource analysis

Description: With this feature you can analyze assets of given file. Also you can detect and extract embedded payloads from malware samples such as AgentTesla, Formbook etc.

Effective Against:

• .NET Executables

Usage: python qu1cksc0pe.py --file suspicious_file --resource

[!NOTE]
Android APK resource scanning was moved into the Android analyzer. Use:
python qu1cksc0pe.py --file app.apk --analyze --report
The JSON report includes resource_scan. Set SC0PE_ANDROID_REPORT_DETAILED=1 to keep more details in the report.

Hash scan

Description: You can check if hash value of the given file is in built-in malware hash database. Also you can scan your directories with this feature.

Usage: python qu1cksc0pe.py --file suspicious_file --hashscan

Folder scan

Supported Arguments:

• --hashscan
• --packer

Usage: python qu1cksc0pe.py --folder FOLDER --hashscan

VirusTotal

Report Contents:

• Threat Categories
• Detections
• CrowdSourced IDS Reports

Usage for --vtFile: python qu1cksc0pe.py --file suspicious_file --vtFile

[!NOTE]
In Web UI flow, Standart Analysis, Document, and Archive presets also execute VirusTotal file lookup in background and show the result in the report page.

Document scan

Description: This feature can perform deep file inspection against given document files. For example: You can detect and extract possible malicious links or embedded exploits/payloads from your suspicious document file easily!

Effective Against:

• Word Documents (.doc, .docm, .docx)
• Excel Documents (.xls, .xlsm, .xlsx)
• Portable Document Format (.pdf)
• OneNote Documents (.one)
• Rich Text Format Documents (.rtf)
• VBScript/VBA Family (.vbs, .vbe, .vba, .vb, .bas, .cls, .frm)

Usage: python qu1cksc0pe.py --file suspicious_document --docs

Batch Script scan (.bat/.cmd)

Description: Analyze Windows Batch scripts for suspicious commands, encoded payload patterns, URLs/domains/IPs, and rule matches.

Usage: python qu1cksc0pe.py --file suspicious_script.bat --analyze --report

Embedded File/Exploit Extraction

Archive File Scan

Description: With this feature you can perform checks for suspicious files against archive files.

Effective Against:

• ZIP
• RAR
• ACE

Usage: python qu1cksc0pe.py --file suspicious_archive_file --archive

With report: python qu1cksc0pe.py --file suspicious_archive_file --archive --report

With AI support: python qu1cksc0pe.py --file suspicious_archive_file --archive --ai

[!NOTE]
ACE archive support requires 7z/7zz to be installed (setup scripts install it on supported systems).

File signature analyzer

Description: With this feature you can detect and extract embedded executable files(.exe, .elf) from given file. Also you can analyze large files (even 1gb or higher) and extract actual malware samples from them (pumped-file analysis).

Usage: python qu1cksc0pe.py --file suspicious_file --sigcheck

File Carving

MITRE ATT&CK Technique Extraction

Description: MITRE ATT&CK mapping is generated automatically during static analysis based on detected APIs/functions/patterns.

Effective Against:

• Windows Executables
• Linux Executables

Usage: python qu1cksc0pe.py --file suspicious_file --analyze

Android Permission Risk Classification

Description: Android permission analysis now classifies permissions as Dangerous, Special, or Info and writes an aggregate permission_summary into JSON reports.

Usage: python qu1cksc0pe.py --file suspicious_app.apk --analyze --report

[!NOTE]
In Web UI report page, Android permissions are shown in a dedicated section with counters and per-permission state badges.

Programming language detection

Description: You can get programming language information from given file.

Usage: python qu1cksc0pe.py --file suspicious_executable --lang

Interactive shell

Description: You can use Qu1cksc0pe in command line mode.

Usage: python qu1cksc0pe.py --console

Dynamic Analysis

Android Application Analysis

[!NOTE]
You must connect a virtual device or physical device to your computer.

Usage: python qu1cksc0pe.py --watch

https://github.com/user-attachments/assets/7b27abb9-f18e-4611-8bdd-cd65106b5cf0

Linux Dynamic Analysis

Description: Linux dynamic flow is menu-driven and lets you choose emulation or live process monitoring.

Usage: python qu1cksc0pe.py --watch

After selecting Linux:

• Option 1: Binary Emulation (isolated environment).
• Option 2: PID Monitoring.

Notes:

• PID monitoring accepts both numeric PID and process name.
• Interactive Linux prompts support TAB autocomplete (selection, path, PID/name).
• Emulation fallback order: Docker SDK -> Docker CLI -> host qemu (if Docker is unavailable).

Windows Process Analysis

Usage: python qu1cksc0pe.py --watch

https://github.com/CYB3RMX/Qu1cksc0pe/assets/42123683/a2c84b8f-c12c-47ac-96e9-c345aeda1f54

References

• The Cyber Security Hub
• Kitploit - Top 20 Most Popular Hacking Tools in 2021
• CSIRT.MAI
• Vulners
• RedPacket Security
• Bournemouth University - CERT
• Hacking Articles - Digital Forensics Tools Mindmap
• HackGit - Twitter Post
• Daily Dark Web - Twitter Post
• SANS ISC - Blog Post
• Korben - Blog Post
• heise online - Blog Post

Star History