Leviath logo

Leviath

Show HN: Leviath – Structured context regions for LLM agents, in one Rust binary

Visit Website

At a glance

Explore the listed capabilities, then check the details that matter to you.

  • Single file agent runtime definition
  • Model selection within runtime configuration
  • Toolset definition for each agent
Related by shared tags and keywordsagent-guardheadless-terminalToolbay Stack
show-hnagentaiai assistantautomation
Leviath screenshot 1

Overview

Leviath is a runtime for structured AI agents, distributed as a single 74 MB Rust binary. It organizes agent execution into stages, each with its own model, toolset, and context budget, rather than using a single loop for the entire task. The system includes context regions that pin critical information like task and plan while allowing conversation to compact, and a run journal that records every tool call so that interrupted runs can resume without repeating completed actions.

Leviath supports thousands of agents within one process and provides live steering to interrupt or redirect agents at any depth. It ships with seven pre-built agents and includes a browser-based interface called The Lair for monitoring runs. The runtime is designed to handle tasks where the data exceeds a single context window, and it allows teams to choose between a single loop or a structured pipeline depending on the work.

Pricing

Detailed plans have not been confirmed in our catalog. Check the official website for current limits and billing terms.

Visit Website

Prices and limits may change. Confirm the currency, billing period, seat minimum and usage caps on the official website.

Use Cases

  • Teams use Leviath to run agents on tasks where the data exceeds a single context window.
  • Teams use Leviath to recover from crashes without repeating completed tool calls.
  • Teams use Leviath to run thousands of agents concurrently on one machine.
  • Teams use Leviath to steer sub-agents directly during a run.
  • Teams use Leviath to reduce costs by assigning cheaper models to simpler stages.
  • Teams use Leviath to ensure accuracy by adding a verification stage.

Features

Single file agent runtime definition

Model selection within runtime configuration

Toolset definition for each agent

Per-stage context budget allocation

Thousands of agents in one binary

Compact 39 MB binary

Reviews

No reviews yet. Be the first to share your experience!