The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests external data through subcommands such as sip018-sign and nostr-sign-event. This data is strictly treated as input for cryptographic functions (hashing and signing) and is not executed or interpolated into prompts within the tool itself. The implementation uses robust JSON parsing and hex validation to handle these inputs.
[CREDENTIALS_UNSAFE]: While the skill interacts with sensitive wallet material, it does not hardcode credentials. It appropriately warns against passing passwords as command-line arguments to prevent exposure in process logs and encourages the use of environment variables like AIBTC_WALLET_PASSWORD.
[EXTERNAL_DOWNLOADS]: The skill relies on reputable, industry-standard cryptographic libraries including @stacks/transactions, @noble/curves, and @scure/btc-signer. These dependencies are standard for blockchain development and are not considered risky in this context.
[COMMAND_EXECUTION]: The code uses Bun.spawnSync exclusively within the test suite (signing.test.ts) to verify CLI functionality. This is a common and safe practice for integration testing and does not impact the security of the runtime skill.

signing