The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local bash scripts scripts/discover-servers.sh and scripts/execute-code.sh to manage marimo notebook sessions and interact with their server registry and HTTP APIs.
[REMOTE_CODE_EXECUTION]: The execute-code.sh script enables the agent to send and execute arbitrary Python code within a marimo kernel session. This is the core intended functionality of the skill, providing a programmatic interface to the notebook runtime.
[DATA_EXFILTRATION]: The skill supports connecting to non-local marimo servers via the --url parameter in execute-code.sh. While the script prints a warning if the host is not local, this capability allows for data and code to be transmitted to external endpoints.
[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of Python packages into the notebook environment using the ctx.packages.add() API and references JavaScript libraries from the esm.sh CDN.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection due to its interactive nature.
Ingestion points: The agent reads existing notebook cell content (via the Read tool and ctx.cells[target].code introspection).
Boundary markers: There are no explicit markers or safety warnings used to isolate cell content from the agent's execution instructions.
Capability inventory: The skill provides full Python execution capabilities via execute-code.sh and can modify local files.
Sanitization: No sanitization or validation of cell content is performed before processing or execution.

marimo-pair