Skip to content

NaZelo documentation

The NaZelo family: a sandbox that runs code you do not fully trust, a policy that decides what it may reach, and a signed chain that proves what happened.

Everything here is for someone using it — running a workload, integrating the engine, or verifying evidence it produced. Nothing here depends on anything unpublished.

NaZelo — the engine

NaZelo runs a command under layered isolation and produces signed, verifiable evidence of exactly what ran, how it ran, and why it was allowed.

  • Getting started — from a fresh install to a command in the sandbox, its evidence collected and a token verified
  • Reference — trust levels, presets, network and L7 stream-proxy enforcement, seccomp process governance, capability tokens, evidence events, configuration, the full CLI and Python API, the security model and its limits

NaZelo Trust — the composition

Three engines, one guarantee: PimaTika decides egress from a signed manifest, NaZelo enforces the decision, Litatoli seals the evidence — one command, correlated by one run_id.

Where the guarantee stops

Both products say so in their own words, and both are worth reading before you rely on either. NaZelo enforces at the node; it cannot attest what a guest kernel did to itself. NaZelo Trust correlates three engines; enforced and attested are different claims.

Beside the code

These ship with the source, because they are about building it:

Licences

NaZelo and NaZelo Trust are source-available, not open source: Business Source License 1.1, converting to Apache-2.0 on 2030-07-30. You may read, build, run in production and audit them; you may not offer them to a third party as a hosted service.

The critical-domain packages — nazelo-critical and nazelo-trust-critical, which add simulated protocol peers — are proprietary and documented with the package.

For alternative arrangements: support@nazelo.dev


Questions and issues: https://github.com/theAIstep/nazelo/issues