The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill facilitates the installation of the Foundry toolchain via curl -L https://foundry.paradigm.xyz | bash. This targets the official domain of Paradigm, a well-recognized infrastructure provider in the Web3 space.
[COMMAND_EXECUTION]: The skill utilizes several shell-based utilities and security tools, including find, grep, ripgrep, and forge, to scan source code and execute testing logic.
[EXTERNAL_DOWNLOADS]: It suggests installing slither-analyzer via pip, which is the standard distribution method for the Slither static analysis tool developed by Trail of Bits.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of Solidity source code located in SECSKILL_CONTRACT_DIR. This creates an attack surface where malicious code could attempt to exploit the parsing logic of the analysis tools.
Ingestion points: Contract source files located in the user-provided SECSKILL_CONTRACT_DIR.
Boundary markers: None present in the instructions to separate untrusted code from the agent's context.
Capability inventory: The skill uses rg (ripgrep) for static pattern matching and forge for compiling and executing Solidity test suites.
Sanitization: No explicit sanitization or validation of the ingested source code is performed before analysis.

web3-smart-contract