|
OpenShell
NVIDIA OpenShell is an open-source, secure-by-design runtime that executes autonomous AI agents inside kernel-level sandboxes governed by declarative policy....
|
90 % |
|
OpenShell
NVIDIA OpenShell is an open-source, secure-by-design runtime that executes autonomous AI agents inside kernel-level sandboxes governed by declarative policy....
|
90 % |
|
Spillway — Let AI coding agents loose on code that matters
Spillway wraps Claude Code in an OS-enforced sandbox on your own machine. Your secrets stay unreadable, your API token stays outside the sandbox, and you get...
|
89 % |
|
llmsnitch — Watch what your AI coding agent actually did
Local-only execution tracing for Claude Code. Every tool call your agent makes — including the file edits its permission classifier never sees — recorded to...
|
88 % |
|
See What Your AI Tools Actually Do | Vitund
Let AI agents write code, call APIs, and install packages — without risking your credentials, your network, or your infrastructure. Self-host it and your dat...
|
87 % |
|
SafeYolo: Sandboxed coding agents with controlled network and service access
SafeYolo runs Claude Code, OpenAI Codex and other coding agents in isolated Linux sandboxes with controlled network, service and credential access.
|
86 % |
|
Clawker — Local, Self-Hosted Sandbox for Any AI Coding Agent
Local, self-hosted sandbox for any AI coding agent. Run Claude Code, OpenAI Codex, OpenCode, or Pi locally in isolated Docker containers behind a deny-by-def...
|
86 % |
|
OpenTrApp: A safer way to run autonomous CLI agents on your own computer
OpenTrApp wraps autonomous CLI agents in a five-container sandbox on your own machine and lets you chat with them from Telegram. Open-source, MIT-licensed, n...
|
86 % |
|
Ambit — Secure your AI coding agents
Centralized policy and human-in-the-loop control for Claude Code, Codex, Cursor and Pi. The control plane above every agent sandbox.
|
86 % |
|
CageClaw — Sandbox your AI coding agent
CageClaw wraps AI coding agents in a security sandbox with network isolation, domain allowlists, and full traffic monitoring. Open source.
|
85 % |
|
LabGate - Policy-controlled sandboxes for AI coding agents
LabGate wraps AI coding agents like Claude Code and Codex in isolated sandboxes with scoped filesystems, credential blocking, network policies, and audit tra...
|
85 % |
|
Void-Box
Run AI agents without trusting them. Void-Box runs each one in its own microVM with explicit capabilities, no shared kernel, and inspectable traces.
|
85 % |
|
Mutiqo | Runtime Visibility for AI-Assisted Development
Mutiqo gives AI-assisted development teams local runtime visibility into what coding agents are actually doing on developer machines.
|
85 % |
|
Kastellan — the trustworthy personal AI agent
A personal, always-on AI agent built so that security is its foundational property. Every tool in its own kernel sandbox; every plan reviewed before it runs....
|
85 % |
|
sandboxd — run AI coding agents in disposable cloud sandboxes
Give Claude Code, Codex, Cursor and other AI agents full root inside a throwaway, hardware-isolated cloud sandbox — never your own laptop. Each sandbox boots...
|
85 % |
|
Runvault — the secure agent shell
An agent with a real Linux shell — and the secrets locked out. Credentials stay behind a server-side proxy the sandbox never sees, and every action is logged.
|
85 % |
|
Security guardrails for coding agents · agentjail
Offline policy guardrails for coding agents. Block dangerous tool calls, sandbox execution, and audit agent activity. Open source, Apache-2.0.
|
85 % |
|
Ambient Authority: enterprise AI security, independently reported
Independent, vendor-neutral reporting on enterprise AI security: prompt injection, agent security, model supply chain, and AI governance.
|
85 % |
|
ai-jail: An OS sandbox for AI coding agents
ai-jail runs Claude Code, Codex and other AI coding agents inside an OS sandbox on Linux and macOS. The project stays writable. Your keys, tokens and home di...
|
85 % |
|
envpod — Zero-Trust Governance for AI Agents
Kernel-level governance for autonomous AI agents. Every action isolated, auditable, and reversible. OWASP 10/10. One binary, one YAML.
|
84 % |
Detected social media networks