Role: Senior Agentic UI/UX Architect\n# Context: You are designing a next-generation interface specifically for Gemini 2.5, where the AI is the primary operator of the software (Agent-First). The core concept is "Visual Anchors"—persistent UI components that serve as ground truth for the model's reasoning and state management.\n\n# Task: Develop a comprehensive blueprint for an Agentic UI system.\n\n# Framework Requirements:\n1. Define the Anchor Taxonomy: Categorize UI elements into 'Static' (permanent context), 'Dynamic' (state-driven), and 'Volatile' (transient thought traces). Explain how Gemini 2.5 should perceive these via vision or structured data.\n2. Context Persistence: Design a method for the Agent to 'pin' visual anchors into its long-context window to prevent state drift during multi-step tasks.\n3. Bidirectional Feedback Loop: Create a schema where Agent actions (e.g., tool calls) trigger visual transitions that provide immediate feedback to the human observer without interrupting the Agent's flow.\n4. Gemini 2.5 Optimization: Specifically address how to leverage Gemini's multimodal capabilities to 'read' the UI state directly from a rendered frame vs. a JSON tree.\n\n# Deliverables:\n- A high-level architectural diagram (in Mermaid.js).\n- A JSON specification for the Visual Anchor components.\n- A list of 'Interaction Heuristics' for Human-Agent-UI co-habitation.