The Agent Skills Directory

[SAFE]: The skill enforces a mandatory 'Three-Step Confirmation SOP' (Standard Operating Procedure) comprising trade pair confirmation, quote display, and final sign/approval gates. This ensures that no blockchain transactions are executed without explicit user validation.
[SAFE]: Advanced risk assessment logic is implemented to prominently warn users when the exchange value difference or slippage exceeds 5 percent, mitigating risks associated with low liquidity and potential MEV (sandwich) attacks.
[SAFE]: Indirect prompt injection surfaces are addressed through structured data handling and the requirement for human-in-the-loop confirmation before sensitive capabilities are exercised.
Ingestion points: External data returned from tx.quote and tx.swap_detail tool calls.
Boundary markers: Mandatory confirmation tables that isolate external data for user review.
Capability inventory: Execution of financial swaps via the tx.swap tool.
Sanitization: Automated validation of input amounts, slippage ratios, and price impact against safe thresholds.
[SAFE]: Data protection measures are explicitly defined to mask user account IDs and strictly prohibit the plain-text display of sensitive authentication tokens (mcp_token) during agent-user interactions.

gate-dex-mcpswap