The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill is designed to execute arbitrary Python, R, or Julia code within Jupyter kernels using the j-cli exec command as its core functionality.
[COMMAND_EXECUTION]: The skill executes shell commands to launch Jupyter servers (j-cli serve-cmd), install workflow-specific hooks (j-cli setup), and perform notebook format conversions. It utilizes nohup bash -c to manage long-running server processes.
[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the jupyter-jcli package from PyPI via uv tool install, which provides the underlying CLI functionality.
[CREDENTIALS_UNSAFE]: The agent manages authentication to Jupyter servers using the JCLI_JUPYTER_SERVER_TOKEN and JCLI_JUPYTER_SERVER_URL environment variables.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection through its ingestion of external notebook and script content.
Ingestion points: The agent reads and processes .ipynb and .py files during conversion and execution tasks, and ingests outputs returned by Jupyter kernels.
Boundary markers: No specific delimiters or safety instructions are described for the agent when handling content from these files or outputs.
Capability inventory: The skill allows for arbitrary kernel code execution, local file system writes, and the installation of environment-level hooks.
Sanitization: No automated sanitization or filtering of ingested notebook content or kernel execution results is mentioned.

j-cli