skills/denoland/skills/deno-sandbox/Gen Agent Trust Hub

deno-sandbox

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill uses the official Deno package @deno/sandbox via the JSR registry (jsr:@deno/sandbox). This is a legitimate resource maintained by the skill's authoring organization.
  • [COMMAND_EXECUTION]: The skill documents the sandbox.spawn method, which is the primary intended functionality of the SDK. The documentation emphasizes security by demonstrating how to restrict permissions (e.g., using --allow-none) and implement timeouts to prevent denial-of-service or unauthorized access by the untrusted code.
  • [INDIRECT_PROMPT_INJECTION]: The skill provides patterns for processing external data and code (e.g., executeAgentTool). It mitigates risks by recommending full isolation in Firecracker microVMs and performing validation on the execution results (using JSON.parse rather than direct execution).
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 12:01 PM
Security Audit — agent-trust-hub — deno-sandbox