Glossary
Definitions of the terms used across the NeoHive product and docs. Aimed at engineers evaluating context layers for AI coding tools.
- Context engineering
- A context engineering layer sits between your knowledge sources (repos, docs, Jira) and your AI tools. It indexes and understands your codebase, then provides relevant context to AI assistants when they need it. NeoHive implements this via the Model Context Protocol (MCP).
- Model Context Protocol (MCP)
- The open standard for connecting AI tools to external context sources. NeoHive implements this protocol so any MCP-compatible client can consume the context it serves.
- MCP server
- NeoHive runs as an MCP server on your local machine or in your VPC. You add it as a context source in Claude Desktop, Cursor, or any MCP-compatible AI tool by pointing to localhost:3100/mcp.
- Codebase-aware AI
- AI tools that give grounded, codebase-specific responses referencing your actual architecture, internal APIs, and conventions, instead of generic textbook answers. NeoHive makes existing AI tools codebase-aware without replacing them.
- Zero data egress
- No data leaves your infrastructure. NeoHive runs entirely as a local Docker container; all embedding and retrieval happen locally inside the container. It makes zero outbound network calls. On-premises by default and VPC-compatible for regulated environments.
- Context window
- The maximum amount of input a language model can process in a single request. Pasting code into ChatGPT, Cursor, Claude, or Copilot is limited by context window size. NeoHive indexes your entire codebase locally and serves only the relevant slice for each query, sidestepping the limit.
- Retrieval-Augmented Generation (RAG)
- A pattern where a language model is augmented with a retrieval step that fetches relevant external documents at query time, then conditions its answer on them. NeoHive provides the retrieval half of this pattern for code and engineering knowledge: it indexes your repos, docs, and Jira, and serves the relevant slice to MCP-compatible AI tools.
- Generative Engine Optimization (GEO)
- The practice of structuring web content so it is cited correctly by LLM-powered search engines (ChatGPT, Perplexity, Google AI Overviews, Bing Copilot). Differs from classical SEO in that it emphasizes answer-first prose, structured facts, and machine-readable bundles like llms.txt.
See how NeoHive compares to CLAUDE.md, Mem0, and Basic Memory →