The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because user-provided strings, such as wallet names and Miniscript policy templates, are interpolated directly into shell commands without sanitization.
Ingestion points: User inputs for --name and --miniscript-template in SKILL.md.
Boundary markers: None present to distinguish user data from command structure.
Capability inventory: Subprocess execution of the nunchuk CLI and scripts/miniscript.js, and file-write operations for wallet backups.
Sanitization: No evidence of input validation or escaping before command execution.
[COMMAND_EXECUTION]: The skill frequently executes shell commands using the nunchuk CLI tool and a bundled JavaScript helper located at scripts/miniscript.js to compile and analyze wallet policies.
[CREDENTIALS_UNSAFE]: The skill manages highly sensitive cryptographic secrets, including mnemonics, seed phrases, and extended private keys (xprv). It provides instructions to export these secrets into a plain-text file (wallet-backup.txt), which creates a risk of local data exposure.
[EXTERNAL_DOWNLOADS]: The skill references external resources, such as the Bitcoin Core Hardware Wallet Interface (HWI) on GitHub, for managing hardware signers. While the download is not automated by the skill itself, it directs the user to fetch and execute external binaries.

nunchuk-wallet-creation