nolabs-ai /nono

agent runtime security - zero trust, zero setup, zero latency agent sandboxing

★ 4,996⑂ 336👁 17⑃ 1,264● 981▤ 114
RustApache-2.0Created 2026-01-31nono.sh
下载:v0.82.0v0.82.0 ↓0.82.02026-11-14
+7
日增长
4,996
stars
336
forks
17
watchers
1,264
prs
981
issues
114
releases

README

Built by the team that brought you
  Sigstore
  
  The standard for secure software attestation, used by PyPI, npm, brew, and Maven Central


  
  
  
  


  
    
  
   
      
  
  
    
  


---


 [!NOTE]
 In the lead-up to a 1.0 release, APIs are stabilizing. API changes may still occur where necessary, but will be kept to a minimum.

nono is a complete agent runtime security tool. Secure your agents, their tools, their credentials, and the skills and instructions they follow, with low latency, a minimal footprint, and zero setup. Kernel-enforced sandboxing is the foundation for a complete workflow of protection, verification, audit, and recovery.

Bring your agent: Claude Code, Codex, Copilot, OpenCode, Goose, OpenClaw, Hermes, Pi, or your own built with frameworks like ADK, Mastra, Pydantic AI and LangChain. With ease you can onboard your own custom developed agents into nono’s secure environment, and immediately benefit from its comprehensive security features.

nono supports macOS, Linux, and Windows (WSL2).

Unprivileged, and runs anywhere. A single binary, with no root privileges required. Run nono on a developer’s laptop, inside a Kubernetes pod, in a CI workflow, a container, or a microVM. It composes with your existing infrastructure for defence in depth.

 What nono protects

- Tools — Run each tool in its own ephemeral sandbox, isolated from the agent and other tools, with explicit permissions for files, network access, environment variables, and credentials. Control which commands the agent can call and which tools can call one another.
- Credentials — Keep API keys outside the agent’s sandbox with proxy-based injection and accessible to a single tool if need be. Use phantom tokens to restrict credential use to approved tools.
- Network access — Control where agents connect and what they can do there, with domain filtering and API rules for HTTP methods and paths.
- Supply chain and skill attestation — Verify the origin and integrity of skills, hooks, and instruction files with Sigstore. Define which publishers you trust and reject unsigned or tampered files covered by your policy.
- Cryptographic audit — Keep a tamper-evident record of agent sessions and security decisions, with integrity verification and optional signing.
- Rollback and recovery — Review file changes and restore tracked files from verified snapshots when an agent takes a wrong turn.

Your policies stay outside the agent’s control. Define them in composable profiles and enforce rules such as:

- The aws CLI can use a credential, but curl can’t.
- gh can read issues and PRs, but can only post to org/foo.
- git can call ssh under its own policy, while the agent cannot call ssh directly.

---

 Quickstart

 curl

bash
curl -fsSL https://nono.sh/install.sh | sh


 macOS / Linux (Homebrew)
bash
brew install nono


 Nix

The project provides a Nix flake with two outputs:

- default (from source) — builds from source using buildRustPackage (first run compiles the crate

Top issues

Discussions

Latest release

v0.82.0 v0.82.0 2026-11-14
What's Changed
 chore(nix): update prebuilt hashes for v0.81.0 by @github-actions[bot] in https://github.com/nolabs-ai/nono/pull/2230
 feat(cli): record proxy audit events live to audit log by @lukehinds in https://github.com/nolabs-ai/nono/pull/2235
 fix(cli): pause the PTY relay for network approval prompts by @panga in https://github.com/nolabs-ai/nono/pull/2234
 fix(cli): support kitty release and repeat events in detach matcher by @lukehinds in https://github.com/nolabs-ai/nono/pull/2239
 docs(nep): propose nep-0009 for cli oauth login and device enrollment by @lukehinds in https://github

Snapshot: 2026-11-16T09:23:14Z · Data license Apache-2.0 · JSON