The Agent Skills Directory

[COMMAND_EXECUTION]: The script pillar.ts utilizes the child_process.exec module to launch the system's default browser (via open, start, or xdg-open) for the authentication and signing flow.
[CREDENTIALS_UNSAFE]: The skill manages sensitive secp256k1 signing keys in the ~/.aibtc/signing-keys/ directory. The pillar-direct.ts script includes a fallback mechanism for key derivation that uses a hardcoded default string (pillar-direct-default) when the PILLAR_API_KEY environment variable is not set.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external data sources that could be manipulated to influence agent behavior.
Ingestion points: Data is ingested from the Pillar backend API, the Hiro Stacks blockchain API, and the CoinGecko price API.
Boundary markers: The instructions do not define clear delimiters or instruction-bypass warnings for the data retrieved from these external services.
Capability inventory: The skill has the ability to read and write local files (keys and sessions), execute shell commands to open URLs, and make network requests to external APIs.
Sanitization: External API responses are parsed as JSON and interpolated directly into terminal output and agent data structures without specific sanitization against embedded instructions.

pillar