skills/davidondrej/skills/bb-cli/Gen Agent Trust Hub

bb-cli

Pass

Audited by Gen Agent Trust Hub on Sep 6, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill documents the bb plugin install and bb marketplace add commands in SKILL.md, which allow the system to download and execute code from external Git and NPM repositories.
  • [COMMAND_EXECUTION]: The skill documents features like bb automation create --script and bb terminal create in SKILL.md, which enable the execution of shell commands and long-running processes on the host machine.
  • [INDIRECT_PROMPT_INJECTION]: The skill defines ingestion points for external data via commands like bb thread log and bb terminal output. Evidence chain: 1. Ingestion points: thread logs and terminal stdout/stderr. 2. Boundary markers: Child threads inherit parent permission levels and environments. 3. Capability inventory: Commands for file writing, shell execution, and plugin installation. 4. Sanitization: The documentation explicitly warns against inspecting secret files with shell tools like cat or sed.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 6, 2026, 03:32 PM
Security Audit — agent-trust-hub — bb-cli