The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill documentation includes a code snippet for generating a faucet wallet that logs the private key directly to the console (console.log('Private key:', wallet.privateKey)). This practice exposes sensitive credentials to terminal history, logs, or shared environments, which is a significant security risk during setup.
[INDIRECT_PROMPT_INJECTION]: The skill implements automated triggers based on matching substrings within error messages (account not found and insufficient funds). This creates a vulnerability surface where malicious tool outputs or data processed by the agent could mimic these errors to trigger unintended faucet transactions.
Ingestion points: The msg variable used for string matching in SKILL.md is derived from external tool or system outputs.
Boundary markers: The skill does not employ delimiters or source validation to ensure the error messages originate from a trusted blockchain provider.
Capability inventory: The skill possesses the capability to sign and broadcast blockchain transactions using ethers.js as described in SKILL.md.
Sanitization: There is no evidence of sanitization or origin verification for the error strings before they are processed by the logic.

injective-faucet