The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation recommends installation via npx @aibtc/mcp-server@latest --install, which downloads package code from the npm registry.
[COMMAND_EXECUTION]: The skill provides instructions for executing various shell commands for installation, build processes, and wallet management, including npx, npm install, and git clone.
[DATA_EXFILTRATION]: The skill manages sensitive data including BIP39 mnemonics and private keys, which are stored in the ~/.aibtc/ directory. It transmits cryptographic signatures to vendor-managed endpoints at aibtc.com and pillarbtc.com for identity registration, check-ins, and transaction processing.
[PROMPT_INJECTION]: An indirect prompt injection surface is present in the "Active Check-In Loop" where the agent is instructed to fetch and respond to message content from a remote API. \n
Ingestion points: Message content fetched from https://aibtc.com/api/paid-attention as described in references/genesis-lifecycle.md. \n
Boundary markers: No specific delimiters or "ignore previous instructions" safety markers are documented for the processing of external message content. \n
Capability inventory: The skill allows the agent to access local sensitive files (~/.aibtc/), perform network requests, and sign/broadcast blockchain transactions. \n
Sanitization: The documentation does not describe sanitization or validation of the message content retrieved from the external API before the agent is prompted to parse, sign, and respond to it.

aibtc-bitcoin-wallet