mcp-agent vs adk-rust

Python and Rust frameworks for building AI agents

mcp-agent and adk-rust help developers build LLM agents with tools and workflows. mcp-agent centers on Model Context Protocol integrations and Temporal-backed execution, while adk-rust offers a broader Rust framework with provider choice, graph workflows, servers, and realtime voice capabilities.

mcp-agentadk-rust
LanguagePythonRust
LicenseApache-2.0—
Stars8.6k717
Forks893111
Last analyzedOct 3, 2026Oct 21, 2026

Key differences

  • mcp-agent is written in Python and focuses on managing MCP connections and composing agent workflows; adk-rust is written in Rust and provides a modular agent runtime with typed tools and multiple workflow types.
  • mcp-agent offers routers, map-reduce, orchestrator-workers, and evaluator-optimizer patterns; adk-rust also includes graph workflows with checkpoints, subgraphs, and human-in-the-loop interrupts.
  • mcp-agent supports asyncio or Temporal execution, with Temporal for durable workflows; adk-rust describes durable runs, checkpointing, and resume across its workflow and server capabilities.
  • mcp-agent can expose applications as MCP servers and includes a CLI for scaffolding and deployment; adk-rust can serve agents over HTTP with streaming and a runtime UI.
  • mcp-agent uses an Apache-2.0 license and has 8.6k stars; adk-rust has 717 stars, and its license is not included in the supplied facts.
  • adk-rust requires Rust 1.95 or newer and its 3.0.0 release candidate has breaking API changes; mcp-agent users need Python, provider credentials, and MCP configuration, while Temporal durability requires a Temporal runtime and worker setup.

Choose mcp-agent if you…

  • build Python agents that use tools from MCP servers.
  • compose workflows using patterns such as routers or orchestrator-workers.
  • need Temporal-backed pause, resume, and retry behavior, or want to expose an application as an MCP server.
Read the mcp-agent analysis →

Choose adk-rust if you…

  • build agents in Rust with typed tools and a choice of model providers.
  • need graph workflows with checkpointing, durable resume, or human-in-the-loop interrupts.
  • want HTTP serving, realtime voice capabilities, or governance features such as approvals and audit records.
Read the adk-rust analysis →

Compare the numbers side by side, or add more projects

This comparison is generated with AI from the OSRepos analyses of both projects. Always check each project's repository and documentation before choosing.