🎢Phase 2: Marketplace & Servers

With the foundation in place, Phase 2 shifts the focus to user interaction, developer extensibility, and the public launch of Axom AI's two core surfaces: the Axom Playground and the MCP Marketplace. This phase brings Axom AI into the hands of developers, builders, and power users while reinforcing its commitment to openness, extensibility, and voice-native computing.

  1. Public Launch of the MCP dApp Releaseof dApp interface - a voice-native portal that connects users to a growing network of MCP servers. Users will be able to issue natural language commands, select from available tools, and interact with MCP-powered agents in real time via both speech & text.

  2. Core MCPs Go Live on Production A curated set of reference MCPs will be deployed into the production environment, enabling practical interactions out of the box. These include modules for search (Brave), geolocation (Google Maps), file operations (Google Drive), development workflows (GitHub/GitLab), and real-time content (YouTube, Fetch, etc.). Each MCP is configured with scoped permissions and fine-grained execution logic.

  3. Signature-Based Authentication User access will be anchored by cryptographic signatures and on-chain account verification. This enables secure personalization, permissioned access to private MCPs, and decentralized identity integration without introducing centralized logins or API key sprawl.

  4. Third-Party MCP Registration Developers will be able to register and test their own MCPs using the official protocol. A verified registry with metadata, permission schema, and endpoint validation will be provided, allowing any tool to become callable by any Axom agent with minimal configuration.

  5. SDKs and Local Plugins To extend Axom AI’s presence into native developer environments, this phase introduces official SDKs and plugin interfaces that let developers integrate agents with local tools and workflows. Cursor, a coding assistant interface, will feature direct support for invocation and tool browsing. Cline, Axom AI’s command-line shell, will allow users to interact with agents locally using voice or text input - ideal for power users, automation, and toolmakers building outside the browser. These toolchains expand the footprint beyond dApps and into the daily environments where developers build and experiment.

  6. Secure Dev Sandbox & Simulator Axom will introduce a local sandbox environment for custom MCP development. This includes isolated agent runners, call tracing, memory inspection, and secure API proxying—allowing developers to simulate and validate their tools before deploying to the broader ecosystem. Tool calls, authentication scopes, error states, and retry logic can all be tested in real-time against the QEDA runtime.

Last updated