The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes external definitions (sensor criteria, signal definitions, and project names) which could be used as vectors for indirect prompt injection.
Ingestion points: Data is retrieved via list_projects, find_sensors, and find_signals tools.
Boundary markers: The workflow mandates a 'preview and confirm' step where the agent must show the 'normalized configuration' and obtain explicit user consent before committing changes.
Capability inventory: The skill includes write capabilities such as create_sensor, update_sensor, create_signal, and update_signal.
Sanitization: The instructions emphasize presenting 'normalized' previews to the user, though no automated input sanitization of the definitions is described.
[DYNAMIC_EXECUTION]: The skill implements a dynamic tool discovery and invocation pattern to handle environments where primary tools are not directly listed.
The agent is instructed to use search_platform_tools and describe_platform_tool to identify and call tools within the platform-- namespace at runtime.
This logic includes specific constraints to prevent 'guessing' tool availability, requiring the agent to rely only on the results from the platform's description tool.

author-signals-and-sensors