The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell commands to manage the application build and discovery process, including npm run generate, npm run compile, and curl to fetch external resources.
[DYNAMIC_EXECUTION]: The skill implements a runtime code generation workflow where it creates TypeScript source files in the src/umbraco-api/tools/ directory. These files are subsequently integrated into the application via the npm run compile command.
[EXTERNAL_DOWNLOADS]: The skill performs network requests to fetch Swagger/OpenAPI specifications from URLs defined in a local discovery manifest. It explicitly utilizes the -k (or --insecure) flag with curl, which disables SSL certificate verification and exposes the communication to man-in-the-middle (MITM) interception.
[INDIRECT_PROMPT_INJECTION]: The skill ingests external API specifications that directly drive the generation of tool handlers and logic, presenting an attack surface where malicious remote content could influence generated code.
Ingestion points: Fetches remote JSON content from a swaggerUrl defined in .discover.json.
Boundary markers: None identified; the instructions provide templates that are populated directly with data from the external spec.
Capability inventory: Writing TypeScript files to the file system and executing shell commands for compilation.
Sanitization: The instructions do not define validation or sanitization requirements for the remote Swagger content before it is used for code generation.

build-tools