emblem-ai

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly supports custom plugins whose executors call external APIs (e.g., fetchWeatherAPI, nftAPI.getStats in references/plugins.md) and includes Reflexive MCP/CLI modes that discover/connect to external MCP servers and read logs/files (references/reflexive.md), so the agent ingests untrusted third‑party content that can influence tool calls and subsequent actions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly states the EmblemAI SDK connects to api.emblemvault.ai at runtime "for prompt orchestration and tool routing" (https://api.emblemvault.ai), so this external backend directly controls prompts and is a required runtime dependency.

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

• Direct money access detected (high risk: 1.00). The skill explicitly provides wallet-enabled user accounts, wallet provisioning, "transaction signing adapters," and references to trading tools and multi-chain wallet auth (MetaMask, WalletConnect, Phantom, PSBT for Bitcoin). These are specific crypto/blockchain capabilities (wallets and signing) that enable value-moving actions (even if gated by user confirmation), so it grants direct financial execution authority.