Add support for Cowrie personas with patching and startup scripts

- Introduced `patch_cowrie_persona_support.py` to modify Cowrie's source files for persona support, including updates to protocol handling, SSH command responses, netstat output, and service status.
- Created `start-cowrie-persona.py` to manage persona activation, including loading persona configurations, selecting a persona based on environment variables, and starting Cowrie with the selected persona.
- Ensured proper error handling and validation for persona files and configurations.

docker/_tests/README.md

@@ -34,6 +34,8 @@ Individual tests can also be run directly:
 ./docker/_tests/tests/conpot.sh --guardian-ast-port 11001 --ipmi-port 1623
 ./docker/_tests/tests/cowrie.sh
 ./docker/_tests/tests/cowrie.sh --ssh-port 2222 --telnet-port 2323
+./docker/_tests/tests/cowrie.sh --persona debian-bookworm-vuln
+./docker/_tests/tests/cowrie.sh --persona openwrt-1806
 ```
 
 ## Conventions

docker/_tests/tests/cowrie.sh

@@ -11,6 +11,7 @@ DEFAULT_IMAGE="ghcr.io/telekom-security/cowrie:24.04.1"
 IMAGE=""
 SSH_PORT=""
 TELNET_PORT=""
+PERSONA=""
 LOG_DIR=""
 TTY_LOG_DIR=""
 DL_DIR=""
@@ -29,6 +30,7 @@ Options:
   --image IMAGE        Image to test. Defaults to docker/cowrie/docker-compose.yml.
   --ssh-port PORT      Host TCP port for SSH. Default: dynamic free port.
   --telnet-port PORT   Host TCP port for Telnet. Default: dynamic free port.
+  --persona ID         Force a generated Cowrie persona. Default: random.
   --timeout SEC        Timeout for startup, protocol, and log checks. Default: 30.
   --bind-ip IP         Host IP to bind. Default: 127.0.0.1.
   --keep-artifacts     Keep temporary compose file and logs for debugging.
@@ -66,6 +68,15 @@ parse_args() {
         TELNET_PORT="${1#*=}"
         shift
         ;;
+      --persona)
+        [[ $# -ge 2 ]] || test_die "--persona requires an argument"
+        PERSONA="$2"
+        shift 2
+        ;;
+      --persona=*)
+        PERSONA="${1#*=}"
+        shift
+        ;;
       --timeout)
         [[ $# -ge 2 ]] || test_die "--timeout requires an argument"
         TEST_TIMEOUT="$2"
@@ -107,6 +118,9 @@ validate_args() {
   if [[ -n "${TELNET_PORT}" ]]; then
     test_validate_port "${TELNET_PORT}"
   fi
+  if [[ -n "${PERSONA}" && ! "${PERSONA}" =~ ^[a-z0-9][a-z0-9-]*$ ]]; then
+    test_die "--persona must contain only lowercase letters, numbers, and dashes"
+  fi
 }
 
 prepare_cowrie_harness() {
@@ -136,6 +150,8 @@ services:
     ports:
       - "${TEST_BIND_IP}:${SSH_PORT}:22"
       - "${TEST_BIND_IP}:${TELNET_PORT}:23"
+    environment:
+      COWRIE_PERSONA: "${PERSONA}"
     volumes:
       - "${DL_DIR}:/home/cowrie/cowrie/dl"
       - "${KEYS_DIR}:/home/cowrie/cowrie/etc"
@@ -415,12 +431,16 @@ PY
 
 assert_custom_filesystem() {
   docker exec -i "${TEST_CONTAINER_NAME}" python3 - <<'PY'
+import configparser
+import json
 from pathlib import Path
 import sys
 
 root = Path("/home/cowrie/cowrie")
-pickle_path = root / "src" / "cowrie" / "data" / "fs.pickle"
-honeyfs = root / "honeyfs"
+personas_root = root / "personas"
+metadata_path = personas_root / "personas.json"
+selected_path = Path("/tmp/cowrie/persona")
+runtime_config_path = Path("/tmp/cowrie/runtime/cowrie.cfg")
 offenders = []
 
 
@@ -432,50 +452,128 @@ def read_bytes(path):
         sys.exit(1)
 
 
-if not pickle_path.is_file():
-    print(f"Missing Cowrie filesystem pickle: {pickle_path}", file=sys.stderr)
-    sys.exit(1)
-if not honeyfs.is_dir():
-    print(f"Missing Cowrie honeyfs directory: {honeyfs}", file=sys.stderr)
-    sys.exit(1)
-
-if b"phil" in read_bytes(pickle_path).lower():
-    offenders.append(str(pickle_path))
-
-for item in honeyfs.rglob("*"):
-    if "phil" in item.name.lower():
-        offenders.append(str(item))
-        continue
-    if item.is_file() and b"phil" in read_bytes(item).lower():
-        offenders.append(str(item))
-
-if offenders:
-    print("Cowrie filesystem still contains 'phil': " + ", ".join(offenders), file=sys.stderr)
-    sys.exit(1)
-
-pickle_bytes = read_bytes(pickle_path)
-pickle_size = pickle_path.stat().st_size
-passwd = read_bytes(honeyfs / "etc" / "passwd")
-hostname = read_bytes(honeyfs / "etc" / "hostname")
-os_release = read_bytes(honeyfs / "etc" / "os-release")
-
-if pickle_size < 1000000:
-    print(f"fs.pickle is unexpectedly small: {pickle_size} bytes", file=sys.stderr)
-    sys.exit(1)
-
-checks = [
-    (b"ubuntu", pickle_bytes, "fs.pickle does not contain ubuntu"),
-    (b"ubuntu", passwd, "honeyfs /etc/passwd does not contain ubuntu"),
-    (b"srv01", hostname, "honeyfs /etc/hostname does not contain srv01"),
-    (b"Ubuntu 22.04", os_release, "honeyfs /etc/os-release does not describe Ubuntu 22.04"),
-]
-
-for needle, haystack, message in checks:
-    if needle not in haystack:
+def fail(message):
     print(message, file=sys.stderr)
     sys.exit(1)
 
-print("Cowrie filesystem profile validated: ubuntu@srv01 without phil")
+
+def check_forbidden(path):
+    if "phil" in path.name.lower():
+        offenders.append(str(path))
+    if path.is_file():
+        data = read_bytes(path).lower()
+        forbidden = (
+            b"phil",
+            b"ubuntu 22.04",
+            b"2.6.26-2-686",
+            b"2.6.26-19lenny",
+            b"com/ubuntu/upstart",
+            b"dannf@debian.org",
+        )
+        if any(marker in data for marker in forbidden):
+            offenders.append(str(path))
+
+
+if not metadata_path.is_file():
+    fail(f"Missing Cowrie persona metadata: {metadata_path}")
+if not selected_path.is_file():
+    fail(f"Missing selected Cowrie persona file: {selected_path}")
+if not runtime_config_path.is_file():
+    fail(f"Missing runtime Cowrie config: {runtime_config_path}")
+
+personas = json.loads(metadata_path.read_text(encoding="utf-8"))
+if len(personas) != 10:
+    fail(f"Expected 10 Cowrie personas, found {len(personas)}")
+
+ids = {persona["id"]: persona for persona in personas}
+selected = selected_path.read_text(encoding="utf-8").strip()
+if selected not in ids:
+    fail(f"Selected persona is not in personas.json: {selected}")
+
+config = configparser.ConfigParser()
+config.read(runtime_config_path)
+selected_dir = personas_root / selected
+expected_filesystem = selected_dir / "fs.pickle"
+expected_honeyfs = selected_dir / "honeyfs"
+expected_processes = selected_dir / "cmdoutput.json"
+expected_txtcmds = selected_dir / "txtcmds"
+
+if config.get("shell", "filesystem", fallback="") != str(expected_filesystem):
+    fail("Runtime config does not point to the selected fs.pickle")
+if config.get("honeypot", "contents_path", fallback="") != str(expected_honeyfs):
+    fail("Runtime config does not point to the selected honeyfs")
+if config.get("shell", "processes", fallback="") != str(expected_processes):
+    fail("Runtime config does not point to the selected cmdoutput.json")
+if config.get("honeypot", "txtcmds_path", fallback="") != str(expected_txtcmds):
+    fail("Runtime config does not point to the selected txtcmds")
+if config.get("ssh", "version", fallback="") != ids[selected]["ssh_banner"]:
+    fail("Runtime config SSH banner does not match selected persona metadata")
+
+for persona_id, persona in ids.items():
+    persona_dir = personas_root / persona_id
+    pickle_path = persona_dir / "fs.pickle"
+    honeyfs = persona_dir / "honeyfs"
+    config_path = persona_dir / "cowrie.cfg"
+    cmdoutput_path = persona_dir / "cmdoutput.json"
+    txtcmds_path = persona_dir / "txtcmds"
+
+    if not pickle_path.is_file():
+        fail(f"Missing Cowrie persona pickle: {pickle_path}")
+    if not honeyfs.is_dir():
+        fail(f"Missing Cowrie persona honeyfs: {honeyfs}")
+    if not config_path.is_file():
+        fail(f"Missing Cowrie persona config: {config_path}")
+    if not cmdoutput_path.is_file():
+        fail(f"Missing Cowrie persona cmdoutput.json: {cmdoutput_path}")
+    if not txtcmds_path.is_dir():
+        fail(f"Missing Cowrie persona txtcmds: {txtcmds_path}")
+    if pickle_path.stat().st_size < 1000000:
+        fail(f"{persona_id} fs.pickle is unexpectedly small: {pickle_path.stat().st_size} bytes")
+
+    pickle_bytes = read_bytes(pickle_path)
+    if persona["user"].encode("utf-8") not in pickle_bytes:
+        fail(f"{persona_id} fs.pickle does not contain persona user {persona['user']}")
+    if persona["hostname"].encode("utf-8") not in pickle_bytes:
+        fail(f"{persona_id} fs.pickle does not contain persona hostname {persona['hostname']}")
+    if persona["ssh_banner"] not in config_path.read_text(encoding="utf-8"):
+        fail(f"{persona_id} config does not contain persona SSH banner")
+    cmdoutput = json.loads(cmdoutput_path.read_text(encoding="utf-8"))
+    if not cmdoutput.get("command", {}).get("ps"):
+        fail(f"{persona_id} cmdoutput.json has no ps process list")
+    for command_path in (
+        "bin/df",
+        "bin/dmesg",
+        "bin/mount",
+        "bin/ulimit",
+        "usr/bin/lscpu",
+        "usr/bin/nproc",
+        "usr/bin/top",
+    ):
+        if not (txtcmds_path / command_path).is_file():
+            fail(f"{persona_id} txtcmds is missing {command_path}")
+
+    passwd = read_bytes(honeyfs / "etc" / "passwd")
+    hostname = read_bytes(honeyfs / "etc" / "hostname")
+    os_release = read_bytes(honeyfs / "etc" / "os-release")
+    if persona["user"].encode("utf-8") not in passwd:
+        fail(f"{persona_id} honeyfs /etc/passwd does not contain persona user")
+    if persona["hostname"].encode("utf-8") not in hostname:
+        fail(f"{persona_id} honeyfs /etc/hostname does not contain persona hostname")
+    if not os_release.strip():
+        fail(f"{persona_id} honeyfs /etc/os-release is empty")
+
+    check_forbidden(pickle_path)
+    check_forbidden(config_path)
+    check_forbidden(cmdoutput_path)
+    for item in honeyfs.rglob("*"):
+        check_forbidden(item)
+    for item in txtcmds_path.rglob("*"):
+        check_forbidden(item)
+
+if offenders:
+    fail("Cowrie persona filesystem contains forbidden markers: " + ", ".join(offenders))
+
+print(f"Cowrie persona pool validated: {selected} selected, {len(personas)} personas, no phil")
 PY
 }
 

docker/cowrie/Dockerfile

@@ -48,17 +48,19 @@ RUN apk --no-cache -U upgrade && \
     pip3 install --break-system-packages --upgrade --no-cache-dir pip && \
     pip3 install --break-system-packages --no-cache-dir -r requirements.txt && \
     pip3 install --break-system-packages --no-cache-dir -e . && \
+    python3 /root/dist/patch_cowrie_persona_support.py --cowrie-root /home/cowrie/cowrie && \
     python3 /root/dist/generate_cowrie_fs.py --cowrie-root /home/cowrie/cowrie && \
+    install -m 0755 /root/dist/start-cowrie-persona.py /usr/local/bin/start-cowrie-persona && \
 #
 # Setup configs
     setcap cap_net_bind_service=+ep $(readlink -f $(type -P python3)) && \
-    cp /root/dist/cowrie.cfg /home/cowrie/cowrie/cowrie.cfg && \
+    cp /home/cowrie/cowrie/personas/debian-bookworm-vuln/cowrie.cfg /home/cowrie/cowrie/cowrie.cfg && \
     chown cowrie:cowrie -R /home/cowrie/* /usr/lib/$(readlink -f $(type -P python3) | cut -f4 -d"/")/site-packages/twisted/plugins && \
 #
 # Start Cowrie once to prevent dropin.cache errors upon container start caused by read-only filesystem
     su - cowrie -c "export PYTHONPATH=/home/cowrie/cowrie:/home/cowrie/cowrie/src && \
                     cd /home/cowrie/cowrie && \
-                    /usr/bin/twistd --uid=2000 --gid=2000 -y cowrie.tac --pidfile cowrie.pid cowrie &" && \
+                    /usr/bin/twistd --uid=2000 --gid=2000 --pidfile cowrie.pid cowrie &" && \
     sleep 10 && \
     rm -rf /home/cowrie/cowrie/etc && \
 #
@@ -82,4 +84,4 @@ RUN apk --no-cache -U upgrade && \
 ENV PYTHONPATH=/home/cowrie/cowrie:/home/cowrie/cowrie/src
 WORKDIR /home/cowrie/cowrie
 USER cowrie:cowrie
-CMD ["/usr/bin/twistd", "--nodaemon", "-y", "cowrie.tac", "--pidfile", "/tmp/cowrie/cowrie.pid", "cowrie"]
+CMD ["/usr/local/bin/start-cowrie-persona"]

docker/cowrie/dist/patch_cowrie_persona_support.py

@@ -0,0 +1,114 @@
+#!/usr/bin/env python3
+
+import argparse
+from pathlib import Path
+
+
+def replace_once(path, old, new):
+    text = path.read_text(encoding="utf-8")
+    if old not in text:
+        raise RuntimeError(f"Could not find expected text in {path}")
+    path.write_text(text.replace(old, new, 1), encoding="utf-8")
+
+
+def replace_all(path, old, new):
+    text = path.read_text(encoding="utf-8")
+    if old not in text:
+        raise RuntimeError(f"Could not find expected text in {path}")
+    path.write_text(text.replace(old, new), encoding="utf-8")
+
+
+def patch_protocol(cowrie_root):
+    path = cowrie_root / "src" / "cowrie" / "shell" / "protocol.py"
+    replace_once(path, "import traceback\n", "import traceback\nfrom pathlib import Path\n")
+    replace_once(
+        path,
+        """        try:
+            binary_data = read_data_bytes("txtcmds", *path.lstrip("/").split("/"))
+            return self.txtcmd(binary_data)
+        except FileNotFoundError:
+            pass
+""",
+        """        txtcmds_path = CowrieConfig.get("honeypot", "txtcmds_path", fallback="")
+        if txtcmds_path:
+            operator_cmd = Path(txtcmds_path) / path.lstrip("/")
+            if operator_cmd.is_file():
+                return self.txtcmd(operator_cmd.read_bytes())
+
+        try:
+            binary_data = read_data_bytes("txtcmds", *path.lstrip("/").split("/"))
+            return self.txtcmd(binary_data)
+        except FileNotFoundError:
+            pass
+""",
+    )
+
+
+def patch_ssh(cowrie_root):
+    path = cowrie_root / "src" / "cowrie" / "commands" / "ssh.py"
+    replace_once(
+        path,
+        """        self.write(
+            f"Linux {self.protocol.hostname} 2.6.26-2-686 #1 SMP Wed Nov 4 20:45:37 \\
+            UTC 2009 i686\\n"
+        )
+""",
+        """        kernel_version = CowrieConfig.get("shell", "kernel_version", fallback="5.10.0")
+        kernel_build = CowrieConfig.get("shell", "kernel_build_string", fallback="#1 SMP")
+        hardware = CowrieConfig.get("shell", "hardware_platform", fallback="x86_64")
+        operating_system = CowrieConfig.get("shell", "operating_system", fallback="GNU/Linux")
+        self.write(
+            f"Linux {self.protocol.hostname} {kernel_version} {kernel_build} "
+            f"{hardware} {operating_system}\\n"
+        )
+""",
+    )
+
+
+def patch_netstat(cowrie_root):
+    path = cowrie_root / "src" / "cowrie" / "commands" / "netstat.py"
+    replace_all(path, "@/com/ubuntu/upstart", "/run/systemd/private")
+    replace_once(
+        path,
+        "Fred Baumgarten, Alan Cox, Bernd Eckenfels, Phil Blundell, Tuan Hoang and others\\n",
+        "Fred Baumgarten, Alan Cox, Bernd Eckenfels, Tuan Hoang and others\\n",
+    )
+
+
+def patch_service(cowrie_root):
+    path = cowrie_root / "src" / "cowrie" / "commands" / "service.py"
+    text = path.read_text(encoding="utf-8")
+    start = text.index("        output = (\n")
+    end = text.index("        )\n        for line in output:", start) + len("        )\n")
+    replacement = """        output = (
+            "[ + ]  cron",
+            "[ + ]  dbus",
+            "[ + ]  networking",
+            "[ + ]  ssh",
+            "[ + ]  rsyslog",
+            "[ + ]  udev",
+            "[ + ]  systemd-journald",
+            "[ + ]  systemd-networkd",
+            "[ - ]  bluetooth",
+            "[ - ]  cups",
+            "[ - ]  nfs-server",
+            "[ - ]  postfix",
+            "[ - ]  rpcbind",
+        )
+"""
+    path.write_text(text[:start] + replacement + text[end:], encoding="utf-8")
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Patch Cowrie v3 persona command support.")
+    parser.add_argument("--cowrie-root", required=True, type=Path)
+    args = parser.parse_args()
+    cowrie_root = args.cowrie_root.resolve()
+    patch_protocol(cowrie_root)
+    patch_ssh(cowrie_root)
+    patch_netstat(cowrie_root)
+    patch_service(cowrie_root)
+
+
+if __name__ == "__main__":
+    main()

docker/cowrie/dist/start-cowrie-persona.py

@@ -0,0 +1,79 @@
+#!/usr/bin/env python3
+
+import json
+import os
+import random
+import shutil
+import sys
+from pathlib import Path
+
+
+COWRIE_ROOT = Path("/home/cowrie/cowrie")
+PERSONAS_ROOT = COWRIE_ROOT / "personas"
+RUNTIME_ROOT = Path("/tmp/cowrie/runtime")
+SELECTED_PERSONA_FILE = Path("/tmp/cowrie/persona")
+
+
+def load_personas():
+    metadata_path = PERSONAS_ROOT / "personas.json"
+    with metadata_path.open(encoding="utf-8") as handle:
+        personas = json.load(handle)
+    if not isinstance(personas, list) or not personas:
+        raise RuntimeError(f"No personas found in {metadata_path}")
+    return personas
+
+
+def choose_persona(personas):
+    requested = os.environ.get("COWRIE_PERSONA", "").strip()
+    ids = [persona["id"] for persona in personas]
+    if requested and requested != "random":
+        if requested not in ids:
+            raise RuntimeError(
+                f"Unknown COWRIE_PERSONA={requested!r}; expected one of: {', '.join(ids)}"
+            )
+        return requested
+    return random.choice(ids)
+
+
+def activate_persona(persona_id):
+    persona_dir = PERSONAS_ROOT / persona_id
+    config_path = persona_dir / "cowrie.cfg"
+    if not config_path.is_file():
+        raise RuntimeError(f"Persona config not found: {config_path}")
+    if not (persona_dir / "fs.pickle").is_file():
+        raise RuntimeError(f"Persona fs.pickle not found: {persona_dir / 'fs.pickle'}")
+    if not (persona_dir / "honeyfs").is_dir():
+        raise RuntimeError(f"Persona honeyfs not found: {persona_dir / 'honeyfs'}")
+
+    RUNTIME_ROOT.mkdir(parents=True, exist_ok=True)
+    shutil.copy2(config_path, RUNTIME_ROOT / "cowrie.cfg")
+    SELECTED_PERSONA_FILE.write_text(persona_id + "\n", encoding="utf-8")
+
+
+def main():
+    try:
+        personas = load_personas()
+        persona_id = choose_persona(personas)
+        activate_persona(persona_id)
+    except Exception as exc:
+        print(f"Could not activate Cowrie persona: {exc}", file=sys.stderr)
+        return 1
+
+    print(f"Starting Cowrie persona: {persona_id}", flush=True)
+    os.environ.setdefault("PYTHONPATH", f"{COWRIE_ROOT}:{COWRIE_ROOT / 'src'}")
+    os.chdir(RUNTIME_ROOT)
+    os.execv(
+        "/usr/bin/twistd",
+        [
+            "/usr/bin/twistd",
+            "--nodaemon",
+            "--pidfile",
+            "/tmp/cowrie/cowrie.pid",
+            "cowrie",
+        ],
+    )
+    return 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())