The Agent Skills Directory

[SAFE]: The skill provides architecture guidelines and diagnostic tools for RAG systems. All scripts are self-contained and use standard Python libraries without external dependencies or network operations.\n- [SAFE]: No evidence of prompt injection, obfuscation, or persistence mechanisms was found. The skill operates locally on document corpora to generate optimization reports and architectural designs.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes document corpora for chunking and evaluation tasks (ingestion in chunking_optimizer.py and retrieval_evaluator.py). While it handles untrusted data, its capabilities are limited to reading local files and writing JSON result summaries to user-defined paths. Evidence chain: Ingestion points in DocumentCorpus and load_corpus (local directory reads); Boundary markers are absent; Capability inventory is restricted to file reading and report writing; Sanitization is not applicable as the content is not interpolated into executable or remote context.

rag-architect