The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted natural language descriptions from users to generate Amplitude chart configurations, which is a standard surface for indirect prompt injection.
Ingestion points: The agent receives natural language requests from users as specified in the SKILL.md workflow.
Boundary markers: The instructions do not define specific delimiters (e.g., XML tags or triple quotes) for user input, though the use of tool-based interaction rather than direct string interpolation in system prompts provides a degree of isolation.
Capability inventory: The skill uses tools including Amplitude:query_dataset and Amplitude:save_chart_edits to perform actions based on user input.
Sanitization: The skill relies on verification steps, error handling, and logical checks (‘Verify results’, ‘Check similar charts’) rather than automated input sanitization.

create-chart