Skip to main content

Introduction

What is NanoTerm?​

NanoTerm is an Agent Runtime — infrastructure that gives AI agents a secure, persistent Linux environment to work in. Agents can execute commands, read and write files, install dependencies, run tests, and persist state across sessions.

Unlike a simple code interpreter or sandboxed shell, NanoTerm provides:

  • Stateful workspaces that persist across sessions
  • Governance through command policies and network controls
  • Auditability with full command history and replay
  • Identity via scoped API keys and secret management

The Problem​

Today, when you give an AI agent access to a terminal, you face a choice:

Run locally

The agent modifies your machine directly. Dangerous, hard to audit, impossible to isolate between tasks.

Use a disposable sandbox

Each execution starts from scratch. No state persistence, no dependency caching, no continuity between steps.

Neither option works for real-world agent workflows where:

  • Agents iterate over many steps, building on previous work
  • Multiple agents may need to collaborate on the same task
  • Teams need visibility into what agents are doing
  • Security policies must be enforced consistently

The Solution​

NanoTerm sits between your agents and the operating system, providing a governed workspace that is:

  1. Secure by default

    Each workspace runs in an isolated container with configurable resource limits. Network egress is deny-by-default. Commands can be restricted by policy. Secrets are injected via a broker, never exposed directly.

  2. Stateful and resumable

    Workspaces persist their filesystem, installed packages, and working state. Agents can pause and resume work. Snapshots capture point-in-time state for rollback or branching.

  3. Fully auditable

    Every command execution, file change, and API call is logged with timestamps, exit codes, and duration. Teams can review agent actions, replay sessions, and enforce compliance.

Architecture​

Agent / SDK / CLI
       |
       v
  NanoTerm API
       |
       v
  ┌─────────────────────────┐
  │   Workspace Runtime     │
  │  ┌───────────────────┐  │
  │  │  Isolated Container│  │
  │  │  - Terminal / PTY  │  │
  │  │  - Filesystem      │  │
  │  │  - Process mgmt    │  │
  │  └───────────────────┘  │
  │  Policy │ Secrets │ Audit│
  └─────────────────────────┘

Who is it for?​

  • AI engineering teams deploying coding agents (Claude Code, Codex, etc.)
  • Platform teams building internal agent tooling
  • Enterprises that need audit and governance over agent actions
  • Developers building agents that need persistent, safe execution environments
  • Agents that outgrow the local machine — builds, test suites, or long-running background work the agent needs should run on infrastructure sized for them, not on the laptop it was launched from. Your agent keeps momentum, your laptop stays cool.

Get Started​