submit: 2054407797725990912

projects/2054407797725990912/README.md

@@ -0,0 +1,40 @@
+# sandboxed.sh
+
+**Participant ID:** `2054407797725990912`
+**Submitted via:** `xagt-plugin@0.4.0`
+**Submitted at:** 2026-05-14T03:23:43.892Z
+
+## Description
+
+sandboxed.sh is the safe runtime for autonomous on-chain AI agents.
+
+## Why it fits Build with XAgent x OKX
+
+sandboxed.sh is an existing shipped product for running autonomous coding agents
+inside isolated Linux workspaces. For the hackathon, it adds a focused
+read-only OKX security skill integration: agents can pre-flight token, dApp,
+transaction, signature, and approval risk without signing, broadcasting, or
+exposing wallet secrets to the model.
+
+Featured hackathon work:
+
+- OKX integration PR: https://github.com/Th0rgal/sandboxed.sh/pull/431
+- Screenshot: https://raw.githubusercontent.com/Th0rgal/sandboxed.sh/hackathon/okx-integration/screenshots/okx-security-report.png
+- Demo shot list: https://github.com/Th0rgal/sandboxed.sh/blob/hackathon/okx-integration/DEMO.md
+
+The OKX skill is installed through sandboxed.sh's Git-backed Library and synced
+into Claude Code, OpenCode, and Amp mission environments through shared
+injection logic. The entry is intentionally read-only: the safety boundary is
+the product, not a missing feature.
+
+## Repo
+
+https://github.com/Th0rgal/sandboxed.sh
+
+## Deployed URL
+
+https://sandboxed.sh
+
+## Hackathon
+
+Build with XAgent × OKX (May 2026)