The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for installing the zvec and @zvec/zvec packages via standard package managers (pip, npm). These are identified as vendor resources from 'zvec-ai'.
[EXTERNAL_DOWNLOADS]: The documentation references well-known technology services including OpenAI, Dashscope (Qwen), and Jina for embedding and reranking tasks. These are legitimate integrations for a vector database tool.
[DATA_EXFILTRATION]: No unauthorized exfiltration patterns were detected. The skill correctly utilizes placeholders for sensitive credentials (e.g., your-openai-api-key) and guides users toward standard API interactions.
[PROMPT_INJECTION]: The skill documents the construction of RAG (Retrieval-Augmented Generation) systems, which inherently involve an indirect prompt injection surface when processing external data.
Ingestion points: Document content and search queries processed in rag-system/python.md and data-operations/python.md.
Boundary markers: Not explicitly defined in the provided code examples.
Capability inventory: The skill facilitates file system persistence (create_and_open), network-based embedding generation, and database querying.
Sanitization: Not demonstrated in the basic architectural examples provided.
[SAFE]: The skill is a legitimate developer tool assistant. Its instructions match its stated purpose of aiding vector database development and implementation.

zvec