Works with the AI Agents you already use
dogbolt binary decompiler
Decompile and analyze small binaries using 13 multi-engine decompilers like Ghidra and Hex-Rays.
Free
dogbolt binary decompiler
Example session with this skill installed
Decompile the local 'crackme' ELF binary using Ghidra and explain the logic in the check_password function.
- Read your context and instructions
- Compiled the dogbolt binary decompiler
I have analyzed the crackme binary using the Ghidra decompiler. The file size is 1.2MB, which is within the operational limits. The analysis focused on the check_password function to understand its validation logic.
Binary Analysis: crackme (Ghidra)
| Function | Address | Logic Type | Behavior Summary |
|---|---|---|---|
main | 0x004011d0 | Entry Point | Collects user input via fgets and passes it to check_password. |
check_password | 0x00401120 | Validation | Performs a byte-wise XOR against a hardcoded key. |
print_flag | 0x00401250 | Utility | Decrypts and prints a success message if the check passes. |
/* Decompiled check_password function using Ghidra */
bool check_password(char *input) {
size_t input_len;
int i;
char hardcoded_key[8] = {0x6b, 0x61, 0x64, 0x6f, 0x68, 0x61, 0x63, 0x6b};
input_len = strlen(input);
if (input_len != 8) {
return false;
}
for (i = 0; i < 8; i++) {
/* XOR comparison logic detected */
if ((input[i] ^ 0x2A) != hardcoded_key[i]) {
return false;
}
}
return true;
}
Behavioral Summary
The binary implements a simple XOR-based password check. It expects an 8-character input. Each character is XORed with the constant 0x2A (42) and compared against the byte array kadohack. To pass this check, the input must be the result of hardcoded_key[i] ^ 0x2A.
Next steps
- Run a comparison against
hex-raysto verify the XOR constant detection. - Generate a Python script to calculate the valid input string based on the XOR key.
- Inspect the
print_flagfunction to see how the final payload is stored.
Connects securely to your tools. The creator never sees your data.
About this skill
Reverse-engineering binaries manually is a slow, error-prone process that requires switching between heavy, expensive desktop environments. This skill enables an AI agent to analyze executables and recover C source code using 13 industry-standard decompiler engines via the Dogbolt MCP server.
What it does
- Multi-Engine Decompilation recovers source code using Ghidra, Hex-Rays, Binary Ninja, Angr, and others.
- Cross-Engine Comparison identifies discrepancies between decompiler outputs to resolve obfuscation or logic errors.
- Symbol Analysis traces entry points, exported symbols, and function calls within ELF, PE, and Mach-O files.
- Behavior Mapping identifies string references and system calls to explain high-level binary logic.
How it works
- The agent verifies the local binary file size and architecture.
- The
decompile_binarycommand uploads the file to the Dogbolt service. - The agent selects a target engine like
ghidrafor general analysis orhex-raysfor readability. - The agent parses the returned C code to explain specific routines like crypto logic or main loops.
Frameworks & tools
This skill utilizes the Dogbolt MCP server for cloud-based decompilation and supports Ghidra, Hex-Rays/IDA, Binary Ninja, and Angr.
Why this beats prompting it yourself
Writing prompts to analyze assembly is unreliable. This skill provides structured access to the actual output of professional reverse-engineering tools, ensuring the agent works with real C code rather than hallucinated logic.
Use cases
- Auditing untrusted ELF binaries for malicious system calls.
- Recovering lost source code from legacy Windows .exe files.
- Comparing how different engines handle obfuscated ARM firmware routines.
- Mapping data structures and crypto functions in Mach-O executables.
Known limitations
Supports binaries under 2MB only. Uses a public decompilation service, so it is not suitable for files containing sensitive proprietary secrets or private keys.
Only supports binaries under 2MB. Dogbolt is a public service; do not use for confidential or proprietary code.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 12 days ago
- Free with an Agensi account
Requires
Permissions
Allowed Hosts