The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exposes an attack surface for indirect prompt injection by ingesting untrusted data from the blockchain. \n
Ingestion points: Functions such as alchemy.nft.getNftMetadata, alchemy.core.getLogs, and alchemy.core.getAssetTransfers retrieve data (e.g., NFT names, transaction data, log messages) directly from the blockchain which can be controlled by third parties. \n
Boundary markers: No specific delimiters or safety instructions are provided in the references to treat this data as untrusted. \n
Capability inventory: The skill allows for significant actions such as alchemy.transact.sendTransaction (sending transactions), alchemy.transact.sendPrivateTransaction (Flashbots), and alchemy.notify.createWebhook (creating webhooks targeting arbitrary URLs). \n
Sanitization: The provided documentation does not include examples or requirements for sanitizing or escaping the retrieved blockchain data before it is processed by the agent.
[SAFE]: The documentation describes the use of apiKey and authToken for authenticating with Alchemy services. The examples use placeholders (e.g., 'your-api-key') and correctly instruct the user on where to obtain these credentials, adhering to safe credential management practices.
[SAFE]: The skill references documentation and scripts from the official Alchemy Platform GitHub repository and the 'antfu/skills' repository, which are recognized as legitimate and well-known sources for developer tools.

alchemy