kajogo777/the-agent-sandbox-taxonomy
An open taxonomy and scoring framework for evaluating AI agent sandboxes: 7 defense layers, 7 threat categories, 3 evaluation dimensions, 20+ "sandboxes" scored.
This project provides a common language and framework for evaluating the security of AI agent sandboxes. It helps practitioners understand what goes into securing an AI agent's environment against various threats and what comes out is a clear 'fingerprint' of a sandbox solution's capabilities and limitations, along with guidance on combining tools. AI security engineers, platform teams, or anyone responsible for deploying AI agents securely would use this to assess and select the right safeguarding tools.
Use this if you need to understand, compare, or choose tools that protect your systems from AI agent misbehavior or malicious actions.
Not ideal if you're looking for an implementation guide for specific sandbox technologies or if your primary concern is traditional application security rather than AI agent-specific risks.
Stars
23
Forks
2
Language
Go
License
—
Category
Last pushed
Mar 08, 2026
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/agents/kajogo777/the-agent-sandbox-taxonomy"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Featured in
Higher-rated alternatives
e2b-dev/E2B
Open-source, secure environment with real-world tools for enterprise-grade agents.
alibaba/OpenSandbox
OpenSandbox is a general-purpose sandbox platform for AI applications, offering multi-language...
e2b-dev/infra
Infrastructure that's powering E2B Cloud.
always-further/nono
Secure, kernel-enforced sandbox CLI and SDKs for AI agents. Capability-based isolation with...
boxlite-ai/boxlite
Sandboxes for every agent. Embeddable, stateful, snapshots, and hardware isolation.