The Agent Skills Directory

[COMMAND_EXECUTION]: The skill identifies and executes shell commands stored in the repository's documentation (specifically docs/agents/project.md) to start the application and its dependencies. It also executes the Playwright test runner through the CLI.
[EXTERNAL_DOWNLOADS]: The skill installs the @playwright/test package if an end-to-end testing harness is not already present. Playwright is a standard, well-known testing library maintained by Microsoft.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill documentation recommends capturing and persisting session information (such as a Playwright storageState JSON) in the local repository to facilitate authenticated testing flows without manual login steps.
[INDIRECT_PROMPT_INJECTION]: The skill processes project-specific data from documentation and a ledger to drive its testing logic and command execution.
Ingestion points: Data is read from docs/agents/project.md and an external acceptance ledger slice.
Boundary markers: The instructions do not define specific delimiters or "ignore previous instructions" markers for the ingested data.
Capability inventory: The skill is capable of installing packages, executing shell commands, writing files, and committing code changes.
Sanitization: No specific sanitization or validation of the commands or text read from the repository files is documented.

acceptance-ui