# nolabs-ai/nono

secure multiplexed execution paths for agents - zero trust, zero setup, zero latency.

Repository: https://github.com/nolabs-ai/nono
Canonical: https://ross.abutalabs.com/products/nono
Homepage: https://nono.sh
Language: Rust
License: Apache-2.0
License Family: permissive
Topics: ai-agents, security, ai-agent-security, ai-security, code-execution, mcp, sigstore, supply-chain-security, zero-trust, agent-sandbox, agent-security, ai-security-tool, llm-security, mcp-security, ai-agent-sandbox
Last push: 2026-08-26T15:40:28+00:00

## Health v2 (maintenance only)
Score: 78/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 86, longevity 15
- inputs: {"age_days": 214, "days_push": 7, "days_rel": 14, "gap_med": 1, "n_releases_24m": 106}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3830, forks 252 (observed 2026-08-28T04:08:21.714298+00:00)

## What it is
nono is a Rust-based CLI and SDK that runs AI coding agents and their tool invocations in ephemeral, kernel-enforced micro sandboxes with capability-based least-privilege isolation. It brokers tool execution with argv policy, phantom credentials injected at a proxy boundary, human approval routing, and a hash-chained cryptographic audit trail, with no daemon, container, or VM required.

## Use cases
- sandbox claude code or codex agent tool calls
- run AI agents without exposing real credentials
- enforce least-privilege file and network access for agent commands
- audit what tools an AI agent executed
- require human approval for risky agent actions
- sandbox MCP and LLM workloads
- verify and pin agent tool executables with sigstore

## When to choose
- you run AI coding agents and want kernel-level isolation without containers or VMs
- you need credentials kept out of agent child processes
- you want auditable, policy-driven tool execution with human-in-the-loop approval
- you want zero-setup sandboxing on macOS, Linux, or Windows (WSL2)

## When to avoid
- you need full VM or container-level isolation for untrusted workloads
- you need a GUI or non-CLI workflow
- you need Windows outside WSL2
- you require a fully frozen 1.0 API before pre-1.0 changes

## Facets
- artifact type: cli-tool
- maturity: active
- function: security, cli, developer-tools, cryptography, secrets-management, mcp
- domain: security, developer-tools, large-language-models
- platform: windows, rust, cli, cross-platform
- tags: agent-sandbox, zero-trust, supply-chain-security, sigstore, capability-based-security, landlock, seatbelt, llm-security, tool-execution, audit-chain, phantom-credentials, sandbox, ai-agents, command-line, macos, linux

## Member repositories
- nolabs-ai/nono (main) score 78

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:21.714298+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:26:19.729630+00:00, confidence not recorded.
  - readme: https://github.com/nolabs-ai/nono (fetched 2026-08-28T04:08:21.714298+00:00, sha 847476532af6)
  - homepage: https://nono.sh (fetched 2026-08-29T09:21:02.569810+00:00, sha c163df1fba3b)
  - site_page: https://nono.sh/docs (fetched 2026-08-29T09:21:02.579964+00:00, sha ea27e30fa15c)
  - site_page: https://nono.sh/docs/cli/getting_started/installation (fetched 2026-08-29T09:21:02.582580+00:00, sha 07f41951f2ee)
  - registry_crates: https://crates.io/api/v1/crates/nono (fetched 2026-08-29T09:21:02.584917+00:00, sha 779c93b0e54b)
- Data as of 2026-08-30T08:39:29.467469+00:00.