# Glossary

> Terms used across the Specify product and docs.

Terms commonly used across the Specify product and docs.

| Term | Meaning |
| --- | --- |
| **Workspace** | The unit in which a team shares docs, conversations, decisions, and code history, along with members and connected tools. It's the boundary where data is isolated. |
| **Node** | The core unit of content in a workspace. Documents, folders, and agent rules are all nodes. |
| **Document project** | The unit that creates and maintains developer docs from a GitHub repository. It ties together a folder, sources, skills, and automation settings. |
| **Source** | The GitHub repository and branch a document project uses as evidence. |
| **Document skill** | The writing discipline an agent follows when it generates docs. |
| **Auxiliary skill** | A reference procedure that helps a document skill. You can select up to 5. |
| **Incremental update** | A mode that updates docs by applying only the code changes. Supports automatic updates. |
| **Full regeneration** | A mode that recreates the whole document every time. Doesn't support automatic updates. |
| **Proposed update** | A suggestion to change a document in response to a code change. It includes the evidence for the change and a before-and-after view. |
| **Trigger on push** | A trigger that prepares proposed document updates when code is pushed to a source branch. |
| **Style guide** | Markdown writing rules applied to docs. The project guide takes precedence over the workspace guide. |
| **Inbox** | Where the progress and results of background tasks collect. |
| **Thread** | A unit of conversation with an agent. |
| **Daily briefing** | A daily summary of workspace changes, stale-document alerts, and suggested tasks. |
| **RAG** | Retrieval-augmented generation. An approach where AI searches workspace materials and uses them as evidence before answering or writing a document. See [Workspace RAG](/en/knowledge/rag). |
| **Index** | The process of splitting documents and files into chunks and storing them for RAG search. Indexing runs again when you edit a document or upload a file. |
| **Standards library** | A knowledge source for searching standards such as IEC 62304 and ISO 14971 clause by clause. See [Standards library](/en/knowledge/standards). |
| **[GAP]** | A marker the agent adds to content that isn't supported by a standard or other materials. A person needs to check it. |
| **Reasoning effort** | How deeply a model reasons before answering. Choose from Off, Low, Medium, High, Extra, and Max. |
| **Connector** | An integration that connects an external service such as GitHub, Notion, or Slack so the agent can read from it and work with it. |
| **Plugin** | A package that bundles several skills for installation in one step. Example: medical-device-docs. |
| **Automation** | A routine that has the agent run a prompt on a set schedule. |
| **MCP** | Model Context Protocol. An open standard for connecting AI agents to external tools and data. |
| **Specify MCP** | The MCP server that provides workspace context to external coding agents such as Claude Code and Codex. |
| **MCP token** | A workspace-level token external editors use to authenticate with Specify MCP. Read-only by default. |