OpenViSTAI · Agents · Blockchain · Crypto
Menu
Agent Frameworks

Tinyagents

Tinyagents is a GPL-3.0 Rust framework that combines a provider-neutral model-and-tool agent harness with a durable typed state-graph runtime for resumable and inspectable workflows.

Independent OpenViST project profile, based on linked sources.

Tinyagents — OpenViST project profile in Agent Frameworks.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Tinyagents when a self-hosted Rust service needs provider-neutral agent execution and durable workflow state, especially where resumability, typed state and inspectability are important.

Who it is for

  • Rust developers building model-and-tool agent services
  • Teams designing resumable, checkpointed workflows
  • Developers who want typed state graphs without a Python runtime or framework DSL
  • Builders composing multi-agent work over shared graph, harness and session layers

Documented uses

  • Orchestrate model and tool calls in an agent loop
  • Represent agent workflows with typed nodes, edges and state
  • Checkpoint, resume, replay and inspect graph execution
  • Track sessions, messages, tool calls, cost and run lineage
  • Compose durable multi-agent teams and workflows
Before you choose

Requirements & limits

The repository is public and active, and the project is open source under GPL-3.0. Workspace crates are not published to crates.io; the documented installation uses Git or path dependencies. The project is intended for self-hosted Rust services and exposes CLI and server-oriented platform classifications.

  • Choose the harness for provider-neutral model calls, typed tools, middleware, structured output, streaming, retries, limits, caching and usage or cost accounting.
  • Choose the graph runtime when workflows need typed state, conditional routing, checkpoints, interrupts, fan-out, subgraphs or replay and time travel.
  • Account for host-owned responsibilities: authorization, prompt construction, execution policy and durable-dialect conversion remain outside the framework’s core runtime.
  • Plan dependency management around Git or path references because the workspace crates are not available through crates.io.
  • Use the optional SQLite features where the documented checkpoint or session stores fit the service design.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Add Tinyagents workspace crates to a Rust project

Set up the documented Tinyagents dependencies for building a Rust agent or graph.

Before you start

  • A Rust project
  • Git or path dependency support
  • The Tinyagents repository
  • The separate tinyinference-llm dependency when using the documented message and provider types

Steps

  1. Add the required Tinyagents crates as Git or path dependencies because the workspace crates are not published to crates.io.

  2. Add tinyinference-llm as a separate Git dependency when the code uses its message or provider types.

  3. Select the workspace pieces needed for the implementation, such as tinyagents-harness, tinyagents-graph or tinyagents-registry.

Expected outcome

The Rust project can reference the documented Tinyagents workspace components and its separate inference dependency.

Guide limitations

  • This setup does not by itself configure a model provider or execute an agent workflow.
  • The documented crates are not published to crates.io.
Source-based practical guide

Run the documented basic graph example

Run the repository’s basic typed graph integration example after cloning the project.

Before you start

  • Git
  • Rust and Cargo
  • Access to the Tinyagents repository

Steps

  1. Clone the Tinyagents repository.

  2. Change into the repository directory.

  3. Run cargo run -p tinyagents-integration-tests --example basic_graph.

Expected outcome

Cargo starts the repository’s documented basic_graph integration example.

Guide limitations

  • The source documents the command and example, but does not establish compatibility across environments or guarantee a successful run.
  • This example demonstrates a local typed graph and does not configure a live model provider.
COMPARE OPTIONS

Compare options for Agent orchestration

These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.

ProjectDocumented purposePlatformAccess and requirements
TinyagentsOrchestrate model and tool calls in an agent loopCLI, Linux, macOS, Self-hostedThe repository is public and active, and the project is open source under GPL-3.0. Workspace crates are not published to crates.io; the documented installation uses Git or path dependencies. The project is intended for self-hosted Rust services and exposes CLI and server-oriented platform classifications.
AgeniteDefine typed tools with JSON Schema validation and error handling.CLI
VeldraCreate an agent with a policy, tools, skills, knowledge base and reasoning method.Self-hostedThe documented deployment is self-hosted and uses a sign-in gate after a first-run administrator account is created.
Evidence and transparency

Sources & review

Source record details
Source availability
Active
Sources checked
3 October 2026
Original website title
tinyhumansai/tinyagents · Official homepage title · retained in its source language