mirror of
https://github.com/webcoyote/awesome-AI-sandbox.git
synced 2026-11-18 18:03:41 +00:00
Added YOLO-Jail and OpenShell
This commit is contained in:
@@ -94,6 +94,7 @@ AI coding agents are useful precisely because they can read files, run commands,
|
||||
- [packnplay](https://github.com/obra/packnplay) - Docker-backed command sandbox with worktree and dev-container management.
|
||||
- [sandclaude](https://github.com/binwiederhier/sandclaude) - Opinionated Docker wrapper for Claude Code.
|
||||
- [Sculptor](https://github.com/imbue-ai/sculptor) - Desktop tooling for running agents inside isolated containers and testing changes.
|
||||
- [yolo-jail](https://github.com/mschulkind-oss/yolo-jail) - Nix-defined agent environments run in rootless Podman or Apple Container jails, with optional macOS Seatbelt modes.
|
||||
|
||||
### Linux
|
||||
|
||||
@@ -110,6 +111,7 @@ AI coding agents are useful precisely because they can read files, run commands,
|
||||
- [Cupcake](https://github.com/eqtylab/cupcake) - OPA/Rego-based hook enforcement for coding agents.
|
||||
- [deepclause-sdk](https://github.com/deepclause/deepclause-sdk) - Policy/runtime SDK for DML-style authorization logic.
|
||||
- [nah](https://github.com/manuelschipper/nah) - Deterministic allow/ask/block guard for Claude Code tool calls.
|
||||
- [OpenShell](https://github.com/NVIDIA/OpenShell) - NVIDIA runtime that enforces declarative file, network, and process policies on sandboxed agents, with human review for risky access changes.
|
||||
- [predicate-secure](https://github.com/PredicateSystems/predicate-secure) - Policy-based authorization and post-run verification for agents.
|
||||
- [punkgo-jack](https://github.com/PunkGo/punkgo-jack) - Audit and receipt layer for agent actions via Merkle-logged hook events.
|
||||
- [shannot](https://github.com/corv89/shannot) - Human-in-the-loop execution and approval flow for LLM agents.
|
||||
|
||||
Reference in New Issue
Block a user