Harness Hub
All open source

ZeroClaw

Personal AI assistant runtime shipped as a single Rust binary: own the agent, data and machine.

zeroclaw-labs/zeroclaw

Maintained by ZeroClaw Labs

ZeroClaw is a personal-assistant agent runtime delivered as a single Rust binary that users configure and run on their own hardware. It talks to LLM providers including Anthropic, OpenAI, Ollama and about 20 others. It reaches users through 30+ channels such as Discord, Telegram, Matrix, email, voice, webhooks and a CLI. It acts through tools including the shell, a browser and HTTP. Setup runs through a one-shot quickstart, with documented installs for Linux, macOS, Windows, FreeBSD, NixOS and Docker. Its tagline is 'You own the agent. You own the data. You own the machine it runs on.'

Pros & cons

Each point cites its sources and says what kind of evidence backs it. A second reviewer tried to refute every claim before it was kept.

Pros

  • DocsRust personal-assistant runtime with pluggable providers, fallback chains and routing across Anthropic, OpenAI, Ollama and any OpenAI-compatible API.[1]
  • DocsDefaults to supervised autonomy (medium-risk ops need approval, high-risk ones are blocked), with Landlock/Bubblewrap/Seatbelt/Docker sandboxes.[1]
  • DocsUnusual hardware support (GPIO/I2C/SPI on Raspberry Pi, STM32, ESP32) and event-triggered SOPs with approval gates.[1]
  • MetadataApache-2.0, active releases (v0.8.7 on 2026-10-27), about 33k stars.[1]

Cons

  • User reportsA security-minded HN reviewer found its 'defense-in-depth' focused on chat access control and not prompt-injection or exfiltration (the lethal trifecta).[1][2]
  • User reportsGranular filesystem sandbox policy was still an open RFC.[1]
  • User reportsWindows is second-class: 74 test failures on Windows were reported.[1]
  • User reportsPre-1.0 with heavy internal churn (anti-slop debt and session-persistence trackers), and the repo briefly returned 404 in March 2026.[1][2][3]
Best for
Tinkerers wanting a lightweight, self-hosted assistant on small devices or edge hardware.
Watch out for
Don't assume the sandbox covers prompt-injection exfiltration; keep risky tools behind approvals.

Assessed Oct 29, 2026 · adversarially reviewedSuggest a correction

Featured

openclaw/openclaw

The viral open-source personal AI assistant that really does things.

Maintained by OpenClaw Foundation

4 pros · 4 cons
Featured

deepseek-ai/deepseek-harness

Everything is a Plugin — DeepSeek's open agent harness.

Maintained by DeepSeek

4 pros · 4 cons
Featured

NousResearch/hermes-agent

The agent that grows with you.

Maintained by Nous Research

4 pros · 4 cons

Connections

Relations from the Harness Hub knowledge graph, as charted in the State of Harness reports.

Maintained by · 1

  • ZeroClaw LabsOrganisation

Similar entries · 1

Sources

  1. GitHub repo and README
  2. GitHub API repo metadata