The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill documents the 'Preview.PlaywrightScript' function in 'playwright-advanced.md', which allows for the execution of local C# scripts (.csx files) during the testing process. This capability enables the execution of arbitrary code within the test environment.
[PROMPT_INJECTION]: Multiple files (including 'mda-testing.md', 'playwright-advanced.md', and 'cicd-integration.md') describe an 'AI-Generated Test Pattern' where an AI model interprets Azure DevOps work item criteria to generate Playwright tests. This pattern introduces a surface for indirect prompt injection if the criteria provided in the work items are not validated before being processed by the AI for code generation.
Ingestion points: Acceptance criteria and requirements from Azure DevOps work items (referenced via ADO MCP).
Boundary markers: None identified in the provided descriptions.
Capability inventory: 'pac test run', 'npx playwright test', and 'Preview.PlaywrightAction'.
Sanitization: No sanitization or validation process for the external work item content is described.
[COMMAND_EXECUTION]: The 'cicd-integration.md' file provides pipeline templates that execute shell commands to install developer tools and perform authentication, such as 'Start-Process msiexec.exe' and 'pac auth create'.
[EXTERNAL_DOWNLOADS]: The CI/CD instructions in 'cicd-integration.md' reference downloads of official Microsoft development tools from well-known shortlink services at 'aka.ms/PowerAppsCLI' and 'aka.ms/installazurecliwindows'.
[DATA_EXFILTRATION]: The skill documents best practices for handling sensitive authentication data, such as user emails and pipeline access tokens, using environment variables (e.g., '$env:user1Email' and '$env:SYSTEM_ACCESSTOKEN'). While avoiding hardcoded secrets, these patterns involve the retrieval of sensitive environment data.

test-engine