The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes shell scripts (in SKILL.md) designed to help users set up their environment by bridging credentials from the alchemy CLI to a .env file. These commands are local and intended for developer setup.
[INDIRECT_PROMPT_INJECTION]: Several reference files (e.g., references/data-nft-api.md) acknowledge that data from external sources, such as NFT or token metadata, should be treated as untrusted. The skill provides explicit guidance to sanitize and proxy such content when building applications.
Ingestion points: External metadata from blockchain assets enter the application context through various data APIs (NFT, Token, etc.).
Boundary markers: The instructions recommend developer-side sanitization and the use of environment variables to separate secrets from code.
Capability inventory: The skill uses standard network operations (curl) and local file manipulation (echo, sed) to manage configurations.
Sanitization: Guidance is provided within the recipes to treat external URI and metadata fields as untrusted inputs.

alchemy-api