Goodeye
@Goodeye-Labs
Outcome-aligned AI workflow registry. Design and run AI workflows with verifiers that align an agent to measurable outcomes, with reliable image output.
概要
What is Goodeye?
Goodeye is an outcome-aligned AI workflow registry. It enables designing AI workflows paired with verifiers that score an agent against a measurable result, then running them from an agent or the CLI. Reliable image output is built in.
How to use Goodeye?
Connect to Goodeye’s remote MCP server via Streamable HTTP using the provided JSON configuration (URL https://mcp.goodeye.dev/mcp). Authenticate with OAuth or by passing a good_live_ API key as a Bearer token. Alternatively, install the CLI tool with uv tool install goodeye.
Key features of Goodeye
- Outcome-aligned AI workflow registry with verifiers.
- Score agents against measurable results.
- Run workflows from an agent or CLI.
- Built-in reliable image output.
Use cases of Goodeye
- Designing AI workflows that are automatically scored on outcomes.
- Running verifiable agent tasks from a command-line interface.
- Integrating outcome-based evaluation into existing AI pipelines.
FAQ from Goodeye
How do I authenticate with the Goodeye MCP server?
Sign in with OAuth, or pass a good_live_ API key as a Bearer token in the request.
What transport protocol does Goodeye use?
Goodeye supports remote MCP over Streamable HTTP.
How can I install the Goodeye CLI?
Install it with uv tool install goodeye.
Where can I find the workflow catalog and documentation?
The catalog is at https://goodeye.dev and full documentation is at https://goodeye.dev/docs.
