Jupyter

Enumerate and exploit Jupyter Notebook server instances.

Overview

The jupyter module targets Jupyter Notebook servers, covering server metadata, kernel enumeration, notebook listing and reading, guarded code execution via the WebSocket kernel interface, and proof-of-concept actions for reverse shell capability and pip package installation. POST operations automatically acquire and send _xsrf tokens via cookie jar.

Subcommands

Read-Only (no --force-exploit required)

Subcommand	Description
enum	Server metadata and status
kernels	List active kernels
notebooks	List notebook files; optional --mine-secrets fetches each notebook and scans cells for credentials
read-notebook	Read a notebook file by path

Gated (requires --force-exploit)

Subcommand	Description
exec	Execute code in a running kernel via WebSocket
start-kernel	Start a new kernel on the server
reverse-shell-proof	Prove outbound socket capability via a kernel (uses non-routable TEST-NET address)
pip-proof	Prove pip install capability via a kernel (dry-run only, nothing installed)

Flags

Flag	Required	Description
--target	Yes	Jupyter server URL (e.g., http://127.0.0.1:8888)
--token	No	Jupyter authentication token
--header	No	Custom HTTP headers. Repeatable.
--path	For read-notebook	Notebook file path
--mine-secrets	For notebooks	Fetch every listed notebook and emit findings for embedded secrets (extra API calls)
--kernel	For exec, reverse-shell-proof, pip-proof	Kernel ID to execute in
--code	For exec	Code string to execute

Examples

# Enumerate server
./aipostex jupyter --target http://127.0.0.1:8888 --token demo enum

# List active kernels
./aipostex jupyter --target http://127.0.0.1:8888 --token demo kernels

# List notebooks
./aipostex jupyter --target http://127.0.0.1:8888 --token demo notebooks

# List notebooks and mine cells for API keys / connection strings
./aipostex jupyter --target http://127.0.0.1:8888 --token demo notebooks --mine-secrets

# Read a specific notebook
./aipostex jupyter --target http://127.0.0.1:8888 --token demo \
  read-notebook --path notebooks/analysis.ipynb

# Execute code in a kernel (gated)
./aipostex jupyter --target http://127.0.0.1:8888 --token demo \
  exec --kernel kernel-1 --code "print('hi')" --force-exploit

# Start a new kernel (gated)
./aipostex jupyter --target http://127.0.0.1:8888 --token demo \
  start-kernel --force-exploit

# Prove reverse shell capability (gated, safe — uses TEST-NET)
./aipostex jupyter --target http://127.0.0.1:8888 --token demo \
  reverse-shell-proof --kernel kernel-1 --force-exploit

# Prove pip install capability (gated, dry-run only)
./aipostex jupyter --target http://127.0.0.1:8888 --token demo \
  pip-proof --kernel kernel-1 --force-exploit

Execution Details

The exec command connects to the kernel's WebSocket shell channel and sends an execute_request message. This uses the gorilla/websocket library and supports proxy routing via --proxy (including SOCKS5).

Workflow Progression

discover network (discovers Jupyter on :8888)
  → jupyter enum (server metadata)
    → jupyter kernels (list kernels)
    → jupyter notebooks (list notebook files; add --mine-secrets to scan all cells)
      → jupyter read-notebook --path <path> (read content)
        → jupyter start-kernel (create a kernel, gated)
        → jupyter exec --kernel <id> --code <code> (execute, gated)
          → jupyter reverse-shell-proof --kernel <id> (outbound socket proof, gated)
          → jupyter pip-proof --kernel <id> (pip install proof, gated)