The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external Simulink Requirements (.slreqx) files, which could contain unvalidated text. This is handled by collectSkillContext.m and parseEARSPattern.m.
Ingestion points: Requirement descriptions and summaries are extracted from .slreqx files using the slreq API.
Boundary markers: The skill architecture mandates a manual YAML mapping review in Phase 1. The agent is instructed to wait for explicit user approval before proceeding to Phase 2 (generation).
Capability inventory: The skill uses sltest.harness.create to create test environments and sltest.testsequence APIs to modify test logic. It also uses save_system to write to the local filesystem.
Sanitization: The extractRequirementText function in collectSkillContext.m removes HTML tags from requirement descriptions.
[DYNAMIC_EXECUTION]: The skill dynamically constructs MATLAB code blocks for Simulink Test Sequences by combining requirement phrases with model signal mappings.
Evidence: Configurator scripts such as configureEventDriven.m, configureStateDriven.m, and configureUbiquitous.m use sprintf to assemble MATLAB code strings for the 'Action' field of test steps, which are later executed within the Simulink simulation environment.

simulink-test-sequence-from-ears