The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains command templates that interpolate external data, creating an indirect prompt injection surface.\n
Ingestion points: The MODEL_ID placeholder is used within JSON configuration strings passed to npm run commands in SKILL.md.\n
Boundary markers: There are no boundary markers or delimiters specified to isolate the interpolated model ID from the command structure.\n
Capability inventory: The skill utilizes subprocess execution via npm run for debugging and benchmarking tasks as described in SKILL.md.\n
Sanitization: There is no evidence of sanitization, escaping, or validation for the inputs interpolated into the execution string.\n- [COMMAND_EXECUTION]: The skill provides instructions for the agent to execute local npm scripts such as debug, verify:model, and bench. These commands are necessary for the skill's primary function of diagnosing and benchmarking model inference but constitute a privileged execution capability.

doppler-debug