The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs users to download and execute a shell script from an external, untrusted repository by piping the output of curl directly to bash. This is a highly dangerous pattern as the remote source is not a recognized trusted vendor and could be compromised to execute malicious code on the local machine.
Evidence: Found in SKILL.md and docs/deployment.md: curl -L https://raw.githubusercontent.com/category-labs/foundry/monad/foundryup/install | bash.
[EXTERNAL_DOWNLOADS]: The skill requires downloading several external dependencies and templates from sources outside the trusted vendor list, including specialized blockchain tools and network configurations.
Evidence: Instructions to use forge init --template monad-developers/foundry-monad for project setup.
Evidence: References to multiple Monad-specific RPC endpoints (e.g., https://rpc.monad.xyz) and explorers.
[COMMAND_EXECUTION]: The skill provides a shell script and numerous terminal commands intended to be executed by the agent to manage blockchain interactions, account creation, and contract deployment.
Evidence: templates/deploy-monad.sh automates the deployment process using Foundry commands.
Evidence: Extensive use of forge create, cast wallet, and hardhat ignition commands throughout the examples and documentation.
[PROMPT_INJECTION]: The skill interacts with external blockchain data and explorer APIs, creating a potential surface for indirect prompt injection if the agent processes malicious data returned from these sources without proper validation.
Evidence: Ingestion points include block explorers like MonadVision and various RPC providers listed in SKILL.md.
Capability inventory: The skill has the ability to execute shell commands and perform sensitive transaction signing.
Sanitization: No explicit sanitization logic for external data is provided in the documentation.

monad