The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to run local Python scripts such as evaluate_rag.py and knowledge_gaps.py, as well as use docker exec. These actions are within the expected scope for a tool designed for project evaluation.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to process and judge outputs from a RAG pipeline. Malicious content within retrieved documents or the generated responses could attempt to manipulate the 'LLM-as-a-judge' evaluation results.
Ingestion points: Data is ingested from scripts/golden_dataset.json and the output of the RAG pipeline being evaluated.
Boundary markers: Absent; there are no instructions to wrap external content in delimiters or ignore embedded instructions within the data.
Capability inventory: The skill has the capability to execute shell commands and read files via the Bash and Read tools.
Sanitization: Absent; no validation or sanitization steps are defined for the content before it is processed by the evaluator.

evaluate