The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches the Foundry security toolkit from paradigm.xyz and standard analysis tools via pip and npm. These are well-known services and organizations in the blockchain development ecosystem.
[REMOTE_CODE_EXECUTION]: Instructions in SKILL.md include installing the Foundry toolkit by piping a remote script from paradigm.xyz to bash. This is the standard installation procedure for this well-known development suite.
[COMMAND_EXECUTION]: The skill makes extensive use of CLI tools including cast, forge, slither, and curl to interact with blockchain nodes and analyze contract data.
[INDIRECT_PROMPT_INJECTION]: As an auditing tool, this skill is designed to ingest and analyze untrusted smart contract source code and bytecode, which presents a surface for indirect prompt injection.
Ingestion points: Contract source code and bytecode are retrieved from external RPC nodes and the Etherscan API (SKILL.md).
Boundary markers: No explicit markers are used to isolate untrusted contract data from the agent's instructions.
Capability inventory: The skill utilizes subprocess execution for analysis (cast, forge, slither) and network access for data retrieval (curl).
Sanitization: The instructions do not specify sanitization or escaping protocols for the contract code before it is presented to the model for analysis.

blockchain-auditor