Agent ensembles to design, generate, and select the best code for every task.
No single model wins every task. Run several, keep the best result. Here's why that works.
Install via npm:
npm install -g voratiqRequirements
- Node 20+
- git
- 1+ AI coding agent (Claude >=2.1.63, Codex >=0.107.0, or Gemini >=0.31.0)
- macOS:
ripgrep - Linux (Debian/Ubuntu):
bubblewrap,socat,ripgrep
See the sandbox runtime docs for guidance on the platform-specific dependencies.
Windows is not currently supported.
Get started with a single command:
voratiqThis drops you into an interactive session with an agent of your choice.
That agent gets access to Voratiq tools for orchestration and verification.
Describe what you want built, and the agent can use those tools to:
- turn your request into a structured execution plan
- delegate work to a custom-designed swarm or a multi-stage workflow
- verify outputs and compare alternatives before choosing what to keep
- track outcomes across runs to learn which agents and workflows work best
You keep the native interactive workflow, with more execution depth and control.
Start here:
- Tutorial - End-to-end walkthrough
- CLI Reference - Commands and options
- Core Concepts - Workflow model and design rationale
Configuration:
- Agents - Agent catalog and invocation details
- Orchestration - Which agents run at each stage
- Verification - How agent outputs are checked and scored before selection
- Environment - Runtime dependencies for agents
- Sandbox - Network and filesystem restrictions
Voratiq is available under the MIT License.