The Agent Skills Directory

[Indirect Prompt Injection Surface]: The skill processes external data, including issue URLs and workspace content, to drive UI automation scenarios. * Ingestion points: Data enters the context via the vscode_automation_evidence_start tool (which takes an issue URL and workspace path) and scenario-specific step instructions. * Boundary markers: There are no explicit delimiters to separate untrusted issue content from the agent's instructions. * Capability inventory: The associated automation tools allow the agent to perform mouse/keyboard actions and potentially modify files in the targeted workspace. * Sanitization: No specific sanitization routines are defined, though the skill provides strong instructions to use isolated profiles and disposable workspaces as a risk mitigation strategy.- [Development Environment Setup]: The documentation outlines standard development procedures, such as installing dependencies and downloading runtimes via npm, which are expected for VS Code engineering tasks.

ui-scenario-validation