Repolode
Submit
ldclabs

ldclabs/anda

A Rust framework for building composable AI agent runtimes that combine models, tools, memory and other agents.

View on GitHubWebsite
Stars
449
Forks
55
Open issues
0
Last commit
19 days ago

What is it?

Anda is a Rust framework for building AI agents that combine models, tools, memory, and other agents inside a single runtime. It emphasizes composability, type-safe extension points, asynchronous execution, and runtime control.

The core engine lets developers register agents and tools, route model requests by capability labels such as primary, pro, flash, or lite, call local or remote functions, and isolate context state per agent or tool. Persistence and memory layers are optional and can be added when an application needs them.

The repository is split into a core traits crate, an agent runtime crate, an HTTP server for serving engines, a command-line interface, and a Web3 client for non-TEE environments. Built-in optional extensions cover conversation storage, filesystem and shell access, fetch, notes, todos, MCP integration, and file-backed skills.

Key features

  • 01Agents and tools are registered through stable traits, so components can be combined into larger workflows.
  • 02Model requests are routed by labeled tiers while provider adapters share a common contract.
  • 03CompletionRunner orchestrates multi-turn model calls, tool calls, cancellation and context handoffs.
  • 04Scoped contexts provide isolated state, cache, object storage, HTTP calls and cancellation per agent or tool.
  • 05Optional extensions add memory tools, filesystem and shell access, notes, todos and file-backed skills.
  • 06Tool bundles and MCP servers can be grouped so agents discover and expand only the tools they need.

Use cases

  • Building a Rust application that needs an AI agent calling multiple tools and models.
  • Serving one or more configured agent engines over HTTP with the provided server.
  • Creating a coding agent with shell, filesystem and patch tools under bounded execution.
  • Developing long-running agent sessions that need compact context handoffs and persistent memory.

Things to know

  • The license is listed as Apache-2.0 in the metadata but the README says MIT, so check the LICENSE file before use.

In collections

Open-source AI agent frameworks & appsMCP servers

Topics

#ai-agent#rust#llm#agent-framework#mcp#tool-calling#memory#agent#agi#autonomous

FAQ

What is anda?
A Rust framework for building composable AI agent runtimes that combine models, tools, memory and other agents. Anda is a Rust framework for building AI agents that combine models, tools, memory, and other agents inside a single runtime. It emphasizes composability, type-safe extension points, asynchronous execution, and runtime control.
Is anda free to use? What license does it have?
anda is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is anda still maintained?
Actively maintained. The last commit to anda was on Oct 20, 2026.
What can I use anda for?
Building a Rust application that needs an AI agent calling multiple tools and models. Serving one or more configured agent engines over HTTP with the provided server. Creating a coding agent with shell, filesystem and patch tools under bounded execution. Developing long-running agent sessions that need compact context handoffs and persistent memory.
Where can I get anda?
The source code and installation instructions are on GitHub: https://github.com/ldclabs/anda

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects