The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because its primary purpose is to ingest, render, and analyze external web content and code snippets, which could be manipulated to include malicious instructions.
Ingestion points: The skill ingests untrusted data via target URLs in SKILL.md and code inputs defined in the evals/compatibility-testing.yaml test cases.
Boundary markers: No explicit boundary markers or 'ignore' instructions are defined to separate ingested content from the agent's internal logic.
Capability inventory: The qe-test-executor and qe-visual-tester agents have capabilities for network operations and subprocess execution (e.g., running Playwright tests).
Sanitization: There is no evidence of input sanitization or validation of the remote content before it is processed by the agent.
[SAFE]: The skill appropriately references well-known and trusted third-party services such as BrowserStack, Sauce Labs, and Playwright for cross-platform automation.
[SAFE]: Sensitive information such as API keys is handled securely using environment variables (process.env.BROWSERSTACK_KEY) rather than being hardcoded in the skill files.

compatibility-testing