Skills
Navigate
skills/jimmylv/bibigpt-skill/bibi-vision/Gen Agent Trust Hub

bibi-vision

Fail

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill recommends installing its CLI tool using a remote script piped directly to a shell. Evidence includes instructions in references/auth.md and the scripts/bibi-check.sh utility script. The source URL https://bibigpt.co/install.sh corresponds to the official domain for the BibiGPT service associated with the skill author.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data that could contain malicious instructions designed to manipulate agent behavior.
  • Ingestion points: The videoUrl parameter in visual-analysis.md and the summary text in mindmap.md.
  • Boundary markers: No delimiters or ignore-instructions are used when interpolating these strings into shell commands.
  • Capability inventory: The skill has the ability to execute shell commands through the bibi CLI.
  • Sanitization: No sanitization or validation of input data is implemented before it is passed to the CLI.
  • [COMMAND_EXECUTION]: The skill performs environment checks by executing shell commands to detect the presence and version of the required CLI tool. Evidence: command -v bibi and bibi --version calls found in scripts/bibi-check.sh.
Recommendations
  • HIGH: Downloads and executes remote code from: https://bibigpt.co/install.sh - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Sep 9, 2026, 02:25 AM
Security Audit — agent-trust-hub — bibi-vision