The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests manual test case definitions in markmap and YAML formats to generate automation logic. This presents a vulnerability surface where instructions hidden within test cases could influence agent behavior.
Ingestion points: Manual test cases (markmap) and Schema files mentioned in Workflow One.
Boundary markers: No explicit boundary markers or isolation protocols are defined for the ingested data.
Capability inventory: The skill can write files (.spec.ts) and execute shell commands (npx playwright).
Sanitization: No explicit content sanitization for ingested test cases is described.
[DYNAMIC_EXECUTION]: The skill generates and executes automation scripts at runtime based on the provided inputs.
Evidence: Workflow One describes converting manual cases into executable Playwright specs.
[COMMAND_EXECUTION]: The skill frequently uses the shell to execute the Playwright test runner.
Evidence: SKILL.md and references mention npx playwright test for running and verifying generated automation.
[EXTERNAL_DOWNLOADS]: The skill uses official package managers to fetch standard testing dependencies.
Evidence: references/playwright-conventions.md includes instructions for installing @axe-core/playwright and official browser binaries.

automated-e2e-testing