Flagged: sudo usage. Scanned against the SlowMist agent-security taxonomy, refreshed every 8 hours. Full audit →
by dredozubov · Agent Tool · ★ 173
Last updated: · Indexed by AgentSkillsHub · Auto-synced every 8h
🔒 Is hazmat safe to install? View the security audit →
Hazmat Run AI coding agents with full autonomy on macOS without giving them your real account. User isolation + kernel sandbox + firewall + rollback I built Hazmat because manual approval mode was the worst of both worlds. It broke agent flow, and it still was not a real security boundary. If an agent gets prompt-injected, runs a poisoned dependency, or follows malicious repo instructions, the important question is not whether it asked politely. The important question is what it can reach. Hazmat changes that blast radius. The agent runs as a dedicated macOS user inside OS-level containment, not as you. Start Here If you want the shortest path to a first contained session: That gets you: a dedicated macOS user a per-session seatbelt policy firewall and DNS hardening automatic pre-session snapshot and restore a session contract that tells you exactly what the agent can touch If you want to preview before changing anything, run or . If you use Codex, OpenCode, Gemini, Hermes, Qwen, or Cursor Agent instead of Claude, start with docs/harnesses.md. What a Session Looks Like Every session starts
| Stars | 173 |
| Forks | 11 |
| Language | Go |
| Category | Agent Tool |
| License | MIT |
| Quality Score | 63.1749801541083/100 |
| Last Updated | 2026-09-11 |
| Created | 2026-03-27 |
| Platforms | claude-code, cli, go |
| Est. Tokens | ~18k |
These tools work well together with hazmat for enhanced workflows:
Looking for a hazmat alternative? If you're comparing hazmat with other agent tool tools, these 6 projects are the closest alternatives on Agent Skills Hub — ranked by topic overlap, star count, and community traction.
Give each AI agent its own isolated machine with root, Docker, and systemd. Active defense detects and stops t
Build big projects with Claude. Chief breaks your work into tasks and runs Claude Code in a loop until they're
Daymon puts your favorite AI to work 24/7. It schedules, remembers, and orchestrates your own virtual team. Fr
The Complete AI Development Toolkit for Claude Code. 106 skills, 36 agents, 171 hooks. Install `ork` for stabl
Process manager for AI agents (and humans)
oh-my-zsh for Claude Code — 16 agents, 35 commands, 32 skills, 21 safety hooks in one install. v4.0 adds an ad
Explore other popular agent tool tools:
hazmat is Containment for AI agents - user isolation, sandboxed execution, network controls, backup/rollback. TLA+ verified.. It is categorized as a Agent Tool with 173 GitHub stars.
hazmat is primarily written in Go. It covers topics such as ai-agents, claude-code, cli.
You can find installation instructions and usage details in the hazmat GitHub repository at github.com/dredozubov/hazmat. The project has 173 stars and 11 forks, indicating an active community.
hazmat is released under the MIT license, making it free to use and modify according to the license terms.
The top alternatives to hazmat on Agent Skills Hub include code-on-incus, chief, daymon. Each offers a different approach to the same problem space — compare them side-by-side by stars, quality score, and community activity.
Grades come from a rule-based scan built on the SlowMist agent-security taxonomy, covering 11 red-flag categories including credential harvesting, data exfiltration, and curl | sh installers. It is a first-layer scan, not a manual audit — we say so rather than overstate it.
The scale of the problem is documented independently: Liu et al. (2026), in a study of 31,132 agent skills, report that 26.1% contain security vulnerabilities. Our own full-catalog census is published as a citable open dataset.
Sources & who's responsible: