The Agent Skills Directory

[COMMAND_EXECUTION]: The skill involves running local Python scripts (chunking_optimizer.py, rag_pipeline_designer.py, retrieval_evaluator.py) to analyze document corpora and design architectures. These scripts are well-defined and perform specific technical tasks without invoking arbitrary shell commands.
[DATA_EXFILTRATION]: While the skill reads local files such as documents, queries, and ground truth data for analysis, there are no network operations, API calls, or patterns indicating data exfiltration to external domains.
[REMOTE_CODE_EXECUTION]: No evidence of remote code execution, package installation, or dynamic script loading from external sources was found. The skill relies entirely on the provided source code and the Python standard library.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted document content for analysis and evaluation. However, the processing is limited to structural analysis and statistical calculations (such as TF-IDF and chunking metrics) which do not involve interpreting or executing the content as instructions for the agent.

rag-architect