AI Agent Sandboxing & Safety
20 repos
Tools and frameworks for safely executing, constraining, and monitoring AI agents and code generation systems. This cluster focuses on guardrails, execution isolation, and runtime safety for autonomous AI systems—particularly those that generate or execute code. While most repos are written in Rust (leveraging systems-level safety guarantees) with supporting tools in Python and TypeScript, they converge on a shared problem: how to let AI agents act autonomously while preventing unintended or malicious behavior through confinement, policy enforcement, and sandboxing mechanisms.
Rust · 8
Go · 3
Python · 3
TypeScript · 3
JavaScript · 1
Shell · 1
codex ·2,831
claude-code ·2,684
ai-agents ·2,250
opencode ·2,092
guardrails ·1,971
developer-tools ·1,944
local-first ·1,887
agent-skills ·1,882
codex-plugin ·1,878
ai-coding ·1,878
