The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The scripts/mint.ts script is designed to accept a WIF (Wallet Import Format) private key as a positional command-line argument. This practice is risky as command-line arguments are typically stored in plain text in shell history files (e.g., .bash_history or .zsh_history) and can be visible to other users in process listings.
[DATA_EXPOSURE]: The script transmits the provided private key to a remote endpoint (https://api.1sat.app/1sat/wallet) via the createRemoteWallet function. While this is the standard operating mode for the @1sat/wallet-remote library, it involves sending sensitive credentials to a third-party infrastructure for transaction signing and broadcasting.
[COMMAND_EXECUTION]: The skill utilizes the Bash(bun:*) tool to execute a TypeScript-based minting utility (scripts/mint.ts). This allows the agent to run local code provided with the skill.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by ingesting external data through the metadataJson argument in scripts/mint.ts.
Ingestion points: The metadataJson string is provided as a CLI argument and parsed by JSON.parse in scripts/mint.ts.
Boundary markers: None. The content is parsed and passed directly into the inscription payload.
Capability inventory: The script has the capability to write data permanently to the BSV blockchain via inscribe.execute.
Sanitization: The script performs basic JSON validation but does not filter or sanitize the content for potential downstream instruction injection.

ordinals-create