The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines a clear ingestion surface by reading data from external sources such as blockchain indexers, protocol hooks, and token metadata. It effectively mitigates potential injection risks by including a 'Hard Rule' that explicitly instructs the agent to treat this data as untrusted and never as instructions.\n- [CREDENTIALS_UNSAFE]: The instructions for using external APIs (e.g., Uniswap API, Sugar SDK) adhere to security best practices by recommending the use of environment variables and secret stores for managing API keys and RPC URIs, rather than hardcoding them.\n- [EXTERNAL_DOWNLOADS]: The skill documentation references several official SDKs and tools from well-known organizations. These references are documented as optional, user-managed installation steps for extending functionality and are not automatically or silently executed by the skill itself.

lp-monitor