Report Abuse

Basic Information

Anda is an open source AI agent framework implemented in Rust that integrates with the Internet Computer Protocol (ICP) blockchain and trusted execution environments (TEEs). It is designed to let developers build, run, and compose autonomous, domain-specific agents that can collaborate to solve complex tasks. The repository contains a modular engine, a command line interface, an HTTP server to host multiple engines, storage and vector database integration, a Web3 client SDK, example agents and tooling. Anda emphasizes agents with permanent identities and cryptographic capabilities derived from ICP, secure execution and storage via dTEE, and persistent memory stored both on-chain and in trusted storage so agents can continuously evolve. The project targets both developers who extend the core libraries and engine and non-developers who can use example agents in the agents directory.

Links

Categorization

App Details

Features
Anda emphasizes composability so specialized agents can be combined into cooperative networks. Simplicity is a priority: configuration-driven agents and example implementations lower the barrier for non-developers. Trustworthiness is provided by running agent logic and storage in a decentralized trusted execution environment (dTEE) backed by TEEs, protecting privacy and integrity. Autonomy derives from ICP-based identities and cryptographic capabilities combined with LLM reasoning for decision making. Perpetual memory stores agent states and knowledge on ICP and in trusted storage to enable continuous learning and long-lived agents. The codebase includes an engine, CLI, HTTP server, LanceDB vector storage integration, a Rust Web3 client SDK, tool libraries, example agents such as anda_bot, and character and example directories for rapid prototyping.
Use Cases
Anda helps teams prototype and deploy agent-based applications by providing a reusable runtime, developer tools, examples, and integrations that reduce integration overhead. Developers can extend core engine components, add external service tools, and create new agents within the agents directory. Non-developers can follow provided agent examples to deploy domain agents with configuration. Security-sensitive deployments benefit from TEE and dTEE support while on-chain identity and storage on ICP allow agents to retain persistent identities and memory across executions. Built-in vector storage integration and a Web3 client SDK ease retrieval and blockchain interaction. The HTTP server enables serving multiple engines, making it easier to run and orchestrate many agents in production or experimental networks.

Please fill the required fields*