bybit-trading

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.80). These files instruct fetching and executing a lazily-downloaded JavaScript executable directly from raw.githubusercontent.com (modules/oauth.js) — a high-risk pattern (remote code download + run) even if a checksum is shown.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). SKILL.md’s required auto-update workflow fetches a runtime manifest and then downloads updated module text from Bybit/GitHub raw URLs (outsider-authored free text via public web content), which is then re-read and used by the agent.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly downloads and loads remote manifest/modules at runtime (e.g., https://api.bybit.com/skill/manifest and https://raw.githubusercontent.com/bybit-exchange/skills/main/modules/oauth.js), and even instructs fetching and executing oauth.js, so remote content can control agent behavior and execute code.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). This skill is explicitly a Bybit trading integration: it instructs the agent how to configure API keys/secrets, construct signed requests (HMAC/RSA), and includes concrete POST examples to create orders (place trades), plus rules for executing Mainnet write operations (confirmation flow). Those are specific, explicit financial execution capabilities (placing market/limit orders, transfers/withdrawals, etc.), not generic tooling.