The solver's final answer and the files it wrote, with a list of the tool calls it made.
About
Loads a solver, fills its variables, adds the skills it links to and your brand context, then runs a multi-step AI model loop with tools until the task is done. It returns the final reply, the list of tool calls made, and how many rounds it took.
You give the solver and a request, and optionally variable values, a cap on rounds (30 by default) and an input folder. After a run it records whether the run succeeded as a brand signal and, when files were written, keeps a list of them for that solver.
A run can make many model calls and can call other tools that cost money, so it asks for confirmation before it starts. It currently runs only inside Syvon's agent engine.
- TypeScript
- AI models
Requirements
- Runs with
- Node.js 20 or later
- Operating systems
- windows, macos, linux
- Network
- Sends the prompt, brand context and tool results to the AI model provider over the internet.
Costs
- External costs
- AI model provider: billed to your own key — Not decided until the Vibe is packaged; in Syvon the run uses Syvon model routing and credits, and paid tools the solver calls are charged as well.