The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process tool descriptions, vendor documentation, and configurations from external sources to generate reports.
Ingestion points: The tool_config dictionary used in the evaluate_tool and calculate_total_cost_ownership methods, as well as stakeholder interviews and market research mentioned in the workflow.
Boundary markers: None identified; the instructions do not specify delimiters or "ignore embedded instructions" warnings for the external data being processed.
Capability inventory: The skill uses a Python evaluation framework capable of making network requests (requests.get) and performing data analysis with pandas.
Sanitization: No sanitization, validation, or escaping of input data is described before interpolation into the evaluation logic or reports.
[EXTERNAL_DOWNLOADS]: The Python evaluation framework performs network operations to external endpoints provided in the tool configuration.
Evidence: The _test_performance method in the ToolEvaluator class uses requests.get(api_endpoint) to measure response times and performance metrics.
[DYNAMIC_EXECUTION]: The skill provides a complex Python class template (ToolEvaluator) as a "Technical Deliverable" and instructs the agent to use it for quantitative analysis.
Evidence: A significant Python code block is included in SKILL.md which the agent is expected to execute or adapt to perform its core mission.

testing-tool-evaluator