mintyouragent

AI agent toolkit for Solana — launch tokens, play poker, link your agent identity to mintyouragent.com. Reads agent personality files (SOUL.md) for profile l...

This skill mostly does what it says (token launches, poker, linking), but the 'soul' extraction and 'link' commands will contact mintyouragent.com; the SKILL.md and README don't state exactly what is sent. Before installing or importing any wallet keys: 1) review the mya.py source around the 'soul' and 'link' commands to see which files/fields are POSTed and whether any secret material could be sent; 2) test in a safe environment (use devnet and a throwaway wallet) to observe network traffic or replay requests; 3) avoid importing real recovery keys or private keys until you're confident the code never transmits them (prefer offline signing or export-only flows); 4) if you plan to use 'link', inspect the remote API (mintyouragent.com) and its privacy policy or run the linking step in a controlled sandbox; 5) prefer running with --dry-run or on devnet and review backups stored under ~/.mintyouragent/ to ensure RECOVERY_KEY.txt isn't created in plaintext. If you cannot audit the 'link' implementation, treat the linking/profile features as potentially transmitting sensitive personality/metadata and proceed with caution.

Type: OpenClaw Skill Name: mintyouragent Version: 3.6.3 The skill is classified as suspicious due to several risky capabilities, primarily the ability to override API and RPC endpoints via environment variables (`SOUL_API_URL`, `HELIUS_RPC`, `SOLANA_RPC_URL`) and to disable SSL verification (`SOUL_SSL_VERIFY=false`) as seen in `mya.py` and `SKILL.md`. While the code demonstrates strong security practices like local key signing, input sanitization, path safety validation, and explicit removal of `subprocess` and `ctypes` modules, these configurable network options introduce a vulnerability. If an attacker can compromise the agent's environment variables, they could redirect network traffic to malicious servers, potentially leading to transaction manipulation or data exfiltration, even though the skill itself does not exhibit direct malicious intent or exfiltrate sensitive data by default.