Skills
Navigate
skills/kevinbadi/ai-os-skills/linkedin-message-agent/Gen Agent Trust Hub

linkedin-message-agent

Warn

Audited by Gen Agent Trust Hub on Apr 8, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The script disables TLS certificate validation globally via process.env.NODE_TLS_REJECT_UNAUTHORIZED = "0" and locally for database connections via rejectUnauthorized: false. This bypasses a critical security layer, making all network communications and credentials (such as the DATABASE_URL) vulnerable to interception via Man-in-the-Middle (MITM) attacks. \n- [COMMAND_EXECUTION]: The script uses child_process.execSync to execute shell commands to start and manage browser instances through the clawdbot utility. \n- [DATA_EXFILTRATION]: The skill reads private LinkedIn message content and transmits it to an external database specified by the DATABASE_URL environment variable. \n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection from messages received on LinkedIn. \n
  • Ingestion points: LinkedIn message content is extracted from the DOM in scripts/linkedin-message-agent.js. \n
  • Boundary markers: No delimiters or instructions are used to isolate the untrusted message content from the agent's logic. \n
  • Capability inventory: The agent can automate browser navigation, send messages, and write to a database. \n
  • Sanitization: Inbound message content is evaluated for response generation without any sanitization or validation.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 8, 2026, 08:20 AM
Security Audit — agent-trust-hub — linkedin-message-agent