The Agent Skills Directory

[Indirect Prompt Injection] (HIGH): The skill processes untrusted external data (configuration files) and has high-privilege capabilities.
Ingestion points: Reads flightcontrol.json and flightcontrol.cue files from the user's workspace.
Boundary markers: None. There are no delimiters or instructions to ignore embedded commands within the processed configuration files.
Capability inventory: Modifies local configuration files and executes shell commands via npx flightcontrol-validate.
Sanitization: None. The skill does not validate or sanitize the content of the configuration files before passing them to the validation tool or reflecting them in its own logic.
[Remote Code Execution] (HIGH): The skill explicitly instructs the agent to run npx flightcontrol-validate <config-file>.
Evidence: The Validation section mandates the use of npx, which downloads and executes code from the npm registry at runtime.
Risk: If an attacker can compromise the flightcontrol-validate package or perform a typosquatting attack, they could achieve code execution on the user's machine.
[Command Execution] (MEDIUM): The skill uses subprocess calls to execute validation logic.
Evidence: npx flightcontrol-validate <config-file>.
Risk: While the command itself is parameterized, the underlying execution environment is exposed to the outputs and side effects of the validation tool.

flightcontrol-config