The Agent Skills Directory

Automated Tool Execution: The skill utilizes the gh command-line interface to interact with the microsoft/fluentui repository. It performs read operations to fetch issue details and write operations to apply labels, assignees, and comments. These actions are performed within the scope of the Fluent UI project and require user approval before any changes are committed.
External Browser Validation: The skill includes an optional step to validate bug reproductions using playwright-cli via npx. This involves visiting external URLs provided in issue bodies (e.g., StackBlitz or CodeSandbox). While visiting arbitrary URLs carries inherent risks, the skill mitigates this by using a headless browser environment, focusing on capturing DOM snapshots and console logs, and requiring explicit human confirmation before execution.
Trusted Resource Usage: The skill fetches the @playwright/cli package from a well-known registry. As Playwright is a technology maintained by Microsoft, this dependency represents standard vendor tooling within the development workflow.
Indirect Prompt Injection Surface: The skill processes untrusted data from GitHub issue titles, bodies, and comments. This data is used to generate triage recommendations and draft responses. The risk of the agent following instructions embedded in issue reports is addressed by the structured output format (tables) and the mandatory human review phase before any actions are taken.
Metadata and Documentation Consistency: The skill's instructions, routing tables, and reference files are consistent with established triage guidelines for the Fluent UI project. All referenced GitHub logins and teams are associated with the project's maintenance and ownership structure.

triage-issues