The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill reads external data from markdown files and uses it to generate executable TypeScript code, creating a significant injection surface.
Ingestion points: The skill reads from /workflows/desktop-workflows.md and /workflows/browser-workflows.md (via Phase 1).
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat the content of the workflow files as untrusted data or to ignore embedded instructions.
Capability inventory: The skill has the ability to write files to the local file system and execute shell commands (npm install, npx tsc) via the provided instructions.
Sanitization: The skill lacks explicit sanitization or escaping mechanisms when interpolating user-controlled strings (like actions and verification text) into the generated Playwright test templates.
[COMMAND_EXECUTION]: The skill involves the execution of shell commands to validate the generated output. It instructs the agent to run npm install --ignore-scripts and npx tsc --noEmit within the generated project directory to perform type-checking.
[DYNAMIC_EXECUTION]: The skill dynamically generates a complete software project, including playwright.config.ts, package.json, and multiple test files. This code is generated at runtime based on the interpretation of external markdown documents and is subsequently processed by development tools.

desktop-workflow-to-playwright