Content
# x64dbg-skills
Claude Code plugin providing skills for x64dbg debugger automation.
## Skills
### `/state-snapshot`
Captures a full debuggee state snapshot to disk for offline analysis:
- All committed memory regions as raw binary files
- Complete processor state (registers) as JSON
### `/state-diff`
Compares two state snapshots to identify what changed between two points in time:
- Register changes (instruction pointer advancement, stack movement, flags, etc.)
- Memory region modifications (stack writes, heap mutations, code changes)
- Synthesized narrative explaining what the program did between snapshots
### `/decompile`
Decompiles a function to C-like pseudocode using [angr](https://angr.io/):
- Decompiles the function at the current instruction pointer if no address is specified
- Accepts a specific address or symbol as an argument
- Tries multiple decompiler strategies for best results
- Suggests nearby functions if the specified address isn't a function entry
### `/yara-sigs`
Scans snapshot memory dumps with [YARA](https://virustotal.github.io/yara/) signatures from the [x64dbg yarasigs](https://github.com/x64dbg/yarasigs) database:
- Automatically clones the yarasigs repo (including Yara-Rules and citizenlab submodules) on first use
- Scan categories: **packers & compilers**, **crypto constants**, **anti-debug / anti-VM**, or **all signatures**
- Builds on `/state-snapshot` — uses an existing snapshot or takes a fresh one
- Reports matches grouped by rule with memory region addresses and metadata
### `/tracealyzer`
Traces execution (into or over calls) for N steps or until a condition is met, then analyzes the recorded instruction log:
- Configurable trace mode: step **into** calls or step **over** calls
- Stop on a max instruction count, an x64dbg expression (e.g. `cip == 0x401000`), or both
- Captures a full instruction log to `traces/` with addresses, disassembly, labels, and comments
- Summarizes execution flow, hot spots, API calls, loops, and notable patterns
- Follow-up actions: annotate key addresses in x64dbg, deeper sub-region analysis, deobfuscation
### `/shellcode-analyzer`
Loads, unpacks, and analyzes raw shellcode blobs in x64dbg:
- Launches x64dbg with `timeout.exe` as a sacrificial process (supports 32-bit and 64-bit)
- Allocates memory, writes shellcode, and redirects execution with optional NOP sled
- **Unpacking** — identifies and executes decoder stubs (XOR loops, decompression routines, self-modifying code)
- **Static analysis** — disassembly, YARA scanning (`/yara-sigs`), annotates key addresses with comments and labels
- **Dynamic analysis** — steps through import resolvers, inspects decoded payloads/strings/C2 configs
- Produces annotated shellcode in x64dbg and optional markdown reports
### `/find-oep`
Smart trace-based OEP finder for packed/protected PE executables:
- Traces through packer stubs using intelligent stepping, anti-debug evasion, and heuristic detection (section transitions, stack restoration, compiler entry patterns, IAT population)
- Handles common packers (UPX, ASPack, MPRESS, PECompact, Themida, VMProtect, Enigma) and unknown/custom packers
- Detects and evades anti-debug techniques: PEB flags, timing checks, hardware BP detection, exception tricks, self-checksums
- Leverages `/yara-sigs` for packer identification and `/state-snapshot` for memory capture at OEP
- Leaves the debugger paused at the OEP with a state snapshot for downstream analysis or PE reconstruction
### `/vuln-hunter`
Hunts for vulnerabilities in a running debuggee through systematic analysis:
- **Reconnaissance** — enumerates imports/exports, categorizes I/O functions by attack context (network, file, registry, etc.), and finds cross-references to dangerous sinks
- **Triage** — ranks code paths by attacker reachability and sink severity, presents a prioritized attack surface map
- **Bug hunting** — iteratively analyzes target functions for buffer overflows, integer wraps, format strings, logic flaws; generates test inputs and observes behavior under the debugger
- **PoC development** — builds proof-of-concept Python scripts that demonstrate impact (crash, info leak, code execution)
- Leverages `/decompile` for complex functions and `/tracealyzer` for execution tracing
- Produces annotated targets in x64dbg and optional markdown vulnerability reports
## Prerequisites
- [x64dbg](https://x64dbg.com/) and [x64dbg Automate](https://dariushoule.github.io/x64dbg-automate-pyclient/installation/) installed
- [x64dbg MCP server](https://dariushoule.github.io/x64dbg-automate-pyclient/mcp-server/) configured in Claude Code
- Python 3 with the `x64dbg_automate` pip package installed:
```
pip install x64dbg_automate[mcp] --upgrade
```
- For the `/decompile` skill: [angr](https://pypi.org/project/angr/) (Python >= 3.10):
```
pip install angr
```
- For the `/yara-sigs` skill: [yara-python](https://pypi.org/project/yara-python/) and [Git](https://git-scm.com/):
```
pip install yara-python
```
- For the `/vuln-hunter` skill: [LIEF](https://lief-project.github.io/) for static PE analysis:
```
pip install lief
```
## Installation
Add the marketplace and install the plugin:
```
/plugin marketplace add dariushoule/x64dbg-skills
/plugin install x64dbg-skills
```
## Updating
To update to the latest version:
```
/plugin install x64dbg-skills
```
## Usage
A decent guide that gives good ideas on how to use these skills: [Cooking with x64dbg and MCP](https://x64.ooo/posts/2026-02-12-cooking-with-x64dbg-and-mcp)
## License
MIT
MCP Config
Below is the configuration for this MCP Server. You can copy it directly to Cursor or other MCP clients.
mcp.json
Connection Info
You Might Also Like
everything-claude-code
Complete Claude Code configuration collection - agents, skills, hooks,...
markitdown
MarkItDown-MCP is a lightweight server for converting URIs to Markdown.
servers
Model Context Protocol Servers
Time
A Model Context Protocol server for time and timezone conversions.
Filesystem
Node.js MCP Server for filesystem operations with dynamic access control.
Sequential Thinking
A structured MCP server for dynamic problem-solving and reflective thinking.