Practical overview
When it fits
Use OpenHarness when you want a code-first TypeScript foundation for a custom agent workflow and need to compose execution, tools, sessions or UI integrations.
Who it is for
- TypeScript developers building custom AI agents and tool-using workflows.
- Application developers integrating streaming agent chat into React or Vue interfaces.
Documented uses
- Assemble an agent with a chosen model and tools such as filesystem or shell access.
- Maintain multi-turn conversations with session features including context compaction and retry.
- Compose middleware for tasks such as turn tracking, compaction and retries.
- Connect agent responses to React or Vue streaming chat interfaces.
Requirements & limits
The repository is public and identifies an MIT license. The ChatGPT/Codex OAuth model-provider package is marked experimental.
- The core quick start pairs @openharness/core with the Vercel AI SDK OpenAI provider.
- The repository lists separate React, Vue, virtual-filesystem and ChatGPT/Codex OAuth packages.
- The ChatGPT/Codex OAuth provider is explicitly experimental.
- The README includes CLI, Next.js and Nuxt examples; the examples use OpenAI API-key configuration by default.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Install the core SDK and configure an agent
Set up the documented OpenAI-backed agent example with filesystem and shell tools.
Before you start
- Node.js/npm environment for installing the packages
- An OpenAI API key
- TypeScript project
Steps
Install @openharness/core and @ai-sdk/openai with npm.
Import Agent, filesystem and shell tool helpers and providers from @openharness/core, and the OpenAI provider from @ai-sdk/openai.
Configure an Agent with a name, an OpenAI model, filesystem and shell tools, and a maximum step count.
Call agent.run with a task and consume its text.delta events.
Expected outcome
A configured agent can run a task with the supplied filesystem and shell tools and stream text deltas.
Guide limitations
- The example depends on the configured OpenAI provider and does not document a complete deployment workflow.
- The repository example uses Node filesystem and shell providers.
Set up the repository examples
Install and build the repository before using its CLI, Next.js or Nuxt examples.
Before you start
- Git
- pnpm
- An OpenAI API key for the default example configuration
Steps
Clone the repository and change into its directory.
Add the OpenAI API key to a .env file.
Run pnpm install and pnpm build.
Expected outcome
The repository dependencies are installed and the project is built, preparing the documented examples for their respective run commands.
Guide limitations
- The setup instructions do not include the full configuration or troubleshooting steps for each example.
- The CLI example also documents a --chatgpt option using the experimental ChatGPT/Codex OAuth provider.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 15 October 2026
- Original website title
- MaxGfeller/open-harness · Official homepage title · retained in its source language