abm-diy-campaign

Warn

Audited by Gen Agent Trust Hub on Jul 17, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONCREDENTIALS_UNSAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Downloads and executes the Vibe Prospecting CLI tool using npx @vibeprospecting/vpai@latest during runtime.
  • [COMMAND_EXECUTION]: Provides instructions to install and execute the Playwright browser automation library (npm install -g playwright) and its associated browser binaries.
  • [CREDENTIALS_UNSAFE]: Directs the user and agent to manage sensitive API keys for LinkedIn, Vibe Prospecting, and Figma, and recommends persisting them in environment variables or shell profile files like ~/.bashrc and ~/.zshrc.
  • [DATA_EXFILTRATION]: Includes code to read local configuration data from the path ~/.config/vpai/config.json.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. Data fetched from the Vibe Prospecting API (such as company names or domains) is interpolated into HTML ad templates in references/04-personalize-figma.md using simple string replacement without sanitization. This data is then rendered in a headless browser context (Playwright), creating a risk of HTML/script injection if the source data is compromised or malicious.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 17, 2026, 04:10 PM
Security Audit — agent-trust-hub — abm-diy-campaign