The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches smart contract ABI data from the Etherscan API, which is a well-known and trusted blockchain information provider.\n- [REMOTE_CODE_EXECUTION]: Instructions provide a command to install the 'Foundry' development toolset using curl -L https://foundry.paradigm.xyz | bash. This is a standard installation pattern for a well-known technology provider.\n- [COMMAND_EXECUTION]: Utilizes the cast CLI tool to perform blockchain read calls and fetch contract details. It also generates instructions for using keypo-wallet, which is a tool from the skill's own vendor.\n- [PROMPT_INJECTION]: The skill represents an indirect prompt injection surface because it processes untrusted data (verified ABIs) from the blockchain and uses it to generate a new SKILL.md file. A malicious contract author could craft function names or metadata within a verified ABI to include instructions intended to influence the behavior of the agent using the generated skill.\n
Ingestion points: Contract ABI fetched via Etherscan API and parsed into function lists.\n
Boundary markers: Absent; the generated output template does not include delimiters or instructions to ignore embedded commands in the ABI data.\n
Capability inventory: The skill uses cast interface, cast call, and curl, and writes new skill files to the local file system.\n
Sanitization: No sanitization or validation of function signatures or names retrieved from the ABI is performed before interpolation into the generated skill file.

contract-learner