The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous templates for executing shell commands via npm run. These commands are used to interface with the Doppler CLI for triage, performance benchmarking, and parity checks between Node.js and browser environments.- [PROMPT_INJECTION]: The skill utilizes placeholders like MODEL_ID in shell command templates. This represents a surface for indirect prompt injection or command injection if the agent populates these placeholders using untrusted data without proper sanitization.
Ingestion points: MODEL_ID variable within npm run debug, npm run verify:model, and npm run bench command strings.
Boundary markers: The templates use single quotes for the JSON configuration argument, which provides some containment but does not prevent escape sequences or malicious flag injection if the placeholder itself is compromised.
Capability inventory: Execution of subprocesses and local shell commands through the npm interface.
Sanitization: The instructions provide no guidance or logic for validating, escaping, or sanitizing the content interpolated into the MODEL_ID field.

doppler-debug