← Back to Category

AI Productivity

nova-tools

An open-source toolkit for coordinating AI agents and their infrastructure, with messaging, fleet orchestration, secrets, workload isolation and persistent session memory.

Agent messagingFleet orchestrationTask isolation

Directory information is for reference. Check the official site for current features, pricing and availability. Listing does not imply hands-on testing or a purchase recommendation.

About this product

nova-tools is a collection of independent command-line utilities rather than a single chat application. It separates agent communication, machine inventories, job coordination, encrypted secrets and persistent notes, so operators can start with one utility or assemble a larger fleet workflow.

Key Features

  • nova-bus uses Redis-backed inboxes to exchange messages across models and agent harnesses.
  • nova-fleet orchestrates machines, services and GitHub Actions runners through Ansible and Tailscale.
  • nova-secrets stores encrypted values in Git and exposes only the secrets requested by a command.
  • nova-reserve limits concurrent jobs with file locks; nova-sandbox restricts a command to granted directories.
  • nova-cairn persists session notes as ordinary files and searches them alongside Markdown notes.

Use cases

  • Exchange work messages between agents using different models or terminal environments.
  • Provision services and run GitHub CI jobs across a fleet connected through Tailscale.

Who it is for

  • For developers and technical operators coordinating multiple agents, shell tasks or machines.

Requirements

  • Messaging requires Redis 6.2 or later; configuration inventory uses PostgreSQL; secrets management requires sops, age and Git.
  • A full fleet also requires Ansible, Tailscale and SSH access; these fleet dependencies are not optional.

Quick Start

  1. Read docs/README.md, choose release binaries for your operating system and architecture, and verify them against SHA256SUMS.
  2. Alternatively, build from source with Go as documented, then run the chosen utility with -h to check its executable and options.
  3. Start with the documented single-machine examples and the dependencies of the selected tool; configure the full service stack when moving to a fleet.

Limitations and considerations

  • Platform requirements differ by utility: the sandbox requires Linux 5.13 or later, or macOS.
  • The local demonstration is distinct from a working fleet, which needs external services and machine access to be configured.

License and links

Code license: MIT

Sources

Source review: 2026-11-07