The Agent Skills Directory

[Indirect Prompt Injection Surface]: The skill facilitates the ingestion of project metadata and user-provided diagrams to generate prompts for sub-agents. While this is necessary for its integration logic, it introduces a potential surface where untrusted data could influence agent behavior.
Ingestion points: Processes powerpages.config.json, .datamodel-manifest.json, and user-uploaded images or text descriptions.
Boundary markers: Prompt templates for sub-agents (e.g., the Explore agent) interpolate these data points directly without explicit adversarial delimiters.
Capability inventory: The skill possesses the ability to write files (services, hooks, types), execute bash commands (npm run build), and run specific utility scripts within the plugin environment.
Sanitization: The workflow includes specific steps to validate Dataverse logical names to ensure consistency.
[Local Command Execution]: The skill executes maintenance and build commands as part of the development lifecycle.
Evidence: Uses node to execute scripts located within the ${CLAUDE_PLUGIN_ROOT} for task-specific operations (like creating web roles or table permissions) and runs npm run build to verify the project's state.

integrate-webapi