Roadmap

Inspecto is actively evolving. Our goal is to make it the ultimate bridge between visual UI development and AI agents. Here is a glimpse into what we are planning for the upcoming releases.

Phase 1: Context Enrichment (In Progress)

Now that the automated integration onboarding and core CLI architecture have stabilized, we are focusing on giving AI richer architectural context instead of shipping raw snippets only.

• Props Source Tracking (P0): Track the origin of props so the AI agent knows exactly where data is passed from and where to apply changes.
• Condition / Loop Rendering Context (P0): Provide context on conditionals and loops rendering the component, helping AI understand the structural code logic better.
• Import Chain Resolution (P1): Resolve the import chain to distinguish between internal project components and third-party library components.
• Expression Span (P1): Support precise code replacement utilizing location data.
• Multiple Element Selection: Allow developers to hold a modifier key (e.g., Shift) to select multiple React/Vue components at once and send their combined context to the AI in a single prompt.
• Runtime State & Computed Styles: Beyond static source code, we plan to extract real-time component states (props/state) and final computed CSS styles from the browser, helping AI pinpoint logic and styling bugs more accurately.
• Auto Screenshot Capture: Automatically take a screenshot of the selected component's rendered state in the browser and attach it to the AI prompt alongside the source code.

Phase 2: MCP & Agentic Workflow

We are actively designing Inspecto's Model Context Protocol (MCP) server to establish a standardized, bidirectional communication bridge between AI agents and the browser frontend. Key initiatives include:

• Core State Management: Implement an in-memory store and event bus to track the lifecycle of UI inspections, states, and browser pages.
• HTTP & SSE APIs: Build RESTful APIs for inspection management and Server-Sent Events (SSE) for real-time state synchronization.
• MCP Server & Tools: Provide core MCP tools (e.g., inspecto_watch, inspecto_resolve, inspecto_reply) allowing agents to watch for new UI inspections, read context, and reply to developers.
• Browser Bidirectional UI: Introduce a visual status indicator and an interactive thread panel in the browser, enabling developers to chat directly with the AI agent inspecting the UI element.
• Self-Driving Mode: Enable closed-loop resolution where AI Agents can automatically read errors, fetch source code via MCP, apply fixes, and verify.

Phase 3: Architecture Upgrades

To support multi-agent and real-time workflows, we’re modernizing Inspecto’s transport layer and IDE bridge.

• WebSocket Migration: Primarily resolving the bidirectional communication gap between IDE Extensions and the local DevServer. Moving away from the current one-way URI Scheme invocation, we will establish persistent WebSocket connections, enabling complex instruction routing and real-time state syncing between the IDE and the browser.
  • Agent Journaling: Persist conversations between browser → MCP → IDE so the same context can be replayed later (useful for multi-step tasks).

---

Have a feature request or want to see where each item lands? Track issues labeled roadmap on GitHub Discussions.