The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted analytics query results to generate visual artifacts. This creates a potential surface where malicious instructions embedded in the source data could influence the agent's behavior during the charting process.
Ingestion points: analytics query results processed in SKILL.md.
Boundary markers: The instructions lack explicit delimiters for data interpolation in the prompt context.
Capability inventory: The skill can write files, execute shell commands (pip, python, node), and start a local web server.
Sanitization: The instructions recommend using json.dumps for embedding data in HTML artifacts to prevent escaping errors.
[DYNAMIC_EXECUTION]: The skill generates self-contained HTML/JavaScript artifacts and explicitly instructs the agent to validate the generated code using node --check. This involves the creation and analysis of executable content at runtime.
[EXTERNAL_DOWNLOADS]: The instructions provide commands to install well-known Python libraries (pandas, plotly, kaleido) from standard registries if they are not present in the environment.
[COMMAND_EXECUTION]: The skill utilizes shell commands for environment management, dependency installation, and launching a local development server for viewing interactive charts.

plotting