The Agent Skills Directory

[SAFE]: No malicious behavior, unauthorized data exfiltration, or obfuscation was detected. The skill's functionality is consistent with its stated purpose of performance analysis.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied data, including Project IDs and metric parameters, which are used to generate analysis queries.
Ingestion points: User inputs for core metrics and project identifiers collected in Stage 1 and Stage 2.
Boundary markers: The workflow uses a dialogue-based confirmation process and closing guides for each stage to maintain context.
Capability inventory: The toolset includes analysis adhoc run for data retrieval and analysis report create for persistent reporting.
Sanitization: Instructions mandate the use of structured AI-facing event definitions rather than raw query parameters.
[DYNAMIC_EXECUTION]: The assistant dynamically constructs JSON-formatted query definitions to perform targeted analysis on payment data. This runtime generation is a core functional requirement of the analytical diagnostic process and is executed through verified internal tools.

payment-attribution-analysis