The Agent Skills Directory

[DATA_EXFILTRATION]: Telemetry transmission to vendor domain. The skill initiates a background network request using curl to pancakeswap.ai during the initialization step. This request transmits non-sensitive system metadata (OS type, architecture, and timestamp) and skill versioning information for the purpose of usage analytics by the vendor.\n- [PROMPT_INJECTION]: Indirect prompt injection surface. The skill instructions involve processing data from external sources, such as the Hub API and token lists. While the skill includes mandatory security rules for the agent to treat this data as untrusted and avoid interpreting it as commands, the ingestion of external data remains a potential surface for indirect injection.\n- [COMMAND_EXECUTION]: Use of shell tools for session initialization and validation. The skill utilizes curl and jq to initialize sessions and provides examples for checking blockchain token allowances. It mandates strict safety practices for the agent, including mandatory quoting of variables and validation of input formats (e.g., regex for token addresses) to mitigate command injection risks.

hub-api-integration