arbitrum-stylus

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill includes multiple command examples that pass a private key via a --private-key CLI flag (e.g., --private-key $PRIVATE_KEY), which requires embedding secret values into generated commands/outputs and is therefore insecure.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill includes shell installation commands that fetch-and-execute remote scripts (curl ... | sh / bash) — specifically https://sh.rustup.rs (SKILL.md) and https://foundry.paradigm.xyz (examples/deploy-stylus/README.md) — which download and run remote code during setup.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). The skill is explicitly about deploying and interacting with Stylus (WASM) smart contracts on Arbitrum and includes concrete transaction-level operations: CLI commands that send signed transactions with --private-key (deploy, activate), examples using "cast send" with --private-key, contract methods that transfer tokens (ERC20 transfer/transfer_from, StylusSwapper calling IERC20.transfer), payable deposit functions, and examples showing sending ETH with a call value. These are specific blockchain wallet/transaction actions (signing/sending transfers and deployments), i.e., direct financial execution capability.