Skills
skills/hoangvantuan/claude-plugin/facebook/Gen Agent Trust Hub

facebook

Pass

Audited by Gen Agent Trust Hub on May 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. Ingestion points: The CONTENT parameter in scripts/fb-post.sh which accepts untrusted user input. Boundary markers: None; content is interpolated directly into the automation flow. Capability inventory: Ability to perform browser navigation, text input, and JavaScript execution. Sanitization: No sanitization is applied to the input content.
  • [COMMAND_EXECUTION]: Employs dynamic JavaScript generation and execution via the pinchtab eval command to facilitate image uploads, which allows for powerful browser interactions but increases the potential impact of script misuse.
  • [COMMAND_EXECUTION]: Accesses the local configuration file at ~/.pinchtab/config.json to retrieve authentication tokens and executes Python-based helper scripts (snap-helpers.py, tag-search.py) to process browser state.
Audit Metadata
Risk Level
SAFE
Analyzed
May 5, 2026, 08:41 AM