The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates connection to a remote MCP server hosted at uiverify.ai. This provides the agent with access to visual regression testing tools and data from the UI Verify service.
[COMMAND_EXECUTION]: Documentation provides setup instructions using the claude mcp add command to register the UI Verify MCP server with the agent's environment. This is a standard configuration step for MCP clients.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data including user-submitted review comments and automated AI verdicts.
Ingestion points: Data enters the agent's context through tools like list_comments and get_build via the MCP server.
Boundary markers: The instructions explicitly warn the agent to treat these inputs as data for review rather than instructions to follow, preventing the agent from autonomously obeying embedded commands.
Capability inventory: The agent has the ability to modify project baselines using the accept_build and review_diff tools.
Sanitization: The skill enforces human-in-the-loop guardrails, requiring the agent to categorize and summarize findings for user approval before performing any baseline updates or write operations.

uiverify-triage-visual-changes