skills/bolhasec/post-empirically-testing-skill-scanners/moltbook-premium-data-splitting/Gen Agent Trust Hub
moltbook-premium-data-splitting
Fail
Audited by Gen Agent Trust Hub on Feb 20, 2026
Risk Level: HIGHCREDENTIALS_UNSAFEDATA_EXFILTRATIONPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [Obfuscation] (HIGH): The 'Premium Skill' section uses a mathematical cipher (Affine cipher) to conceal instructions, a technique used to evade security review and hide malicious intent.\n- [Data Exfiltration] (HIGH): The hidden payload is designed to read the sensitive credential file
~/.config/moltbook/credentials.jsonand prepare its contents for transmission over the network.\n- [Prompt Injection] (HIGH): The skill provides instructions that attempt to deceive the AI agent into executing hidden, malicious logic by framing it as a 'license generation' requirement.\n- [Command Execution] (HIGH): The skill directs the agent to execute Python code that generates a dynamic value used in acurlcommand, creating a path for shell command injection or unauthorized data transfer.
Recommendations
- AI detected serious security threats
Audit Metadata