skills/b-open-io/prompts/setup/Gen Agent Trust Hub

setup

Warn

Audited by Gen Agent Trust Hub on Oct 9, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill frequently executes shell commands to perform auditing and installation tasks.
  • scripts/detector.ts uses Bun.spawn to run shell checks defined in plugin manifests.
  • playground/src/app/api/pack/install/route.ts executes npx, claude, and codex commands to install plugins.
  • scripts/playground_server.ts runs bun install and next build during its startup process.
  • [DYNAMIC_EXECUTION]: The auditing engine in scripts/detector.ts dynamically evaluates shell commands stored in manifest.json files found within other plugin directories. This mechanism allows the skill to execute arbitrary code provided by any plugin present in the harness during a system health check. Additionally, the playground launcher performs runtime builds using Next.js.
  • [EXTERNAL_DOWNLOADS]: The skill performs several network operations to vendor and trusted sources.
  • playground/src/app/api/agent-master/update/install/route.ts downloads application updates (DMG files) from bopen.ai.
  • scripts/detector.ts fetches plugin metadata from the bOpen marketplace API.
  • playground/scripts/vendor-audio.ts downloads UI audio assets from a bOpen registry.
  • [INDIRECT_PROMPT_INJECTION]: The skill's auditing functionality represents an indirect injection surface.
  • Ingestion points: The detector reads setup/manifest.json files from all installed plugins (scripts/detector.ts).
  • Boundary markers: Absent; command strings from manifests are passed directly to the shell.
  • Capability inventory: The skill possesses broad system capabilities including subprocess execution (Bun.spawn, execFile) and file system access.
  • Sanitization: No validation or filtering is performed on the command strings read from external manifests.
  • [REMOTE_CODE_EXECUTION]: The skill contains routes and scripts that download and execute external code.
  • The playground launcher (scripts/playground_server.ts) performs a bun install of its own dependencies.
  • The pack/install API route installs external plugins and skills via standard package managers.
  • The self-update mechanism in playground/src/lib/app-update-install.ts downloads and installs a replacement application bundle.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 9, 2026, 03:41 AM