The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's prerequisites involve installing a CLI tool via npx from the author's official NPM package (@mvanhorn/printing-press-library). This is a legitimate installation step for the vendor's provided tooling.
[COMMAND_EXECUTION]: The skill operates by executing the ble-opaque-binary-pp-cli binary. It provides commands for device discovery, status monitoring, and capability inspection, which are performed via system subprocess calls.
[PROMPT_INJECTION]: The skill exhibits a potential surface for indirect prompt injection when the --live flag is used to interact with physical hardware.
Ingestion points: Untrusted data enters the agent context through the output of ble-opaque-binary-pp-cli status, which reads values from BLE characteristics (ab01).
Boundary markers: The output is structured as JSON, but there are no explicit prompt instructions to the agent to ignore potentially malicious content within characteristic values.
Capability inventory: The CLI includes commands for BLE scanning and command execution, though critical actions are withheld by default.
Sanitization: The raw binary data from the device is represented as hex or JSON without additional sanitization filters.
Note: This risk is significantly mitigated by the skill's design, which defaults to a 'Replay' transport that uses simulated data rather than live device interaction.

ble-opaque-binary