Each term links to the page that explains it in full.
An AI node that decides what to do next: call a tool, read the result, and try again until it has an answer. Compare with a chain. See Agents and tools. For agents you chat with, see Agents in the chat.
Branch
Section titled “Branch”One of the paths a run can take after a node such as IF or Switch. Each branch starts at a named output port. See Splitting and branching.
Canvas
Section titled “Canvas”The editor where you add nodes and connect them into a workflow. See Nodes and Connections between nodes.
An AI node that sends your input to the model once and returns the answer, with no tools. Basic LLM Chain, Summarization and Text Classifier are chains. See Agents and tools.
Content Security Policy (CSP)
Section titled “Content Security Policy (CSP)”Rules a website sets that can block some actions from browser extensions on that site. See Browser context.
Credential
Section titled “Credential”A saved login, API key or token that lets a node use an outside service. Credentials are stored with AWFlow and picked in a node’s settings. See Create credentials.
Embedding
Section titled “Embedding”A list of numbers that represents the meaning of a piece of text, so texts with similar meaning can be found together. A knowledge base stores one per passage. See Embeddings and vectors.
Error Trigger
Section titled “Error Trigger”A trigger that starts a workflow when another workflow’s run fails. See Error handling.
Expression
Section titled “Expression”Text between double curly braces, such as {{ $input.title }}, that AWFlow replaces with a value while a node runs. Expressions are sandboxed: they read values and use a fixed set of helper functions, but don’t run JavaScript. See Mapping with expressions.
Grounded answer
Section titled “Grounded answer”An answer whose claims can be traced back to the passages the model was given. RAG aims for grounded answers. See RAG.
Hallucination
Section titled “Hallucination”When a model states something that sounds right but isn’t supported by its input or by the facts. A knowledge base and a clear “not found” fallback reduce it. See RAG.
In-page node
Section titled “In-page node”A node that reads or changes the web page in your active tab: get its text, click a button, fill a form. In-page nodes need the browser extension. See Browser context.
One record passed between nodes, made of named fields, like a row in a table. A node receives a list of items and usually runs once per item. See Items.
Item linking
Section titled “Item linking”How a value read from another step lines up with the item being processed. See Item linking.
Knowledge base
Section titled “Knowledge base”A collection of your documents (files, web pages, tabs, text) cut into passages and searchable by meaning. AI nodes use it through the Local Knowledge node. See Knowledge bases and RAG.
Lambda workflow
Section titled “Lambda workflow”A workflow built to be reused as a single step inside other workflows. See Lambda workflows.
Large language model (LLM)
Section titled “Large language model (LLM)”A model trained on a lot of text that reads and writes language. In AWFlow, you connect one to an AI node as its model. See Models and where they run.
A part of a workflow that repeats, built with the Loop or Split in Batches node. Most nodes already run once per item, so you need a loop less often than you might think. See Looping.
Mapping
Section titled “Mapping”Putting a value from an earlier step into a node’s setting, usually by dragging it from the Input panel. See Mapping in the UI.
Memory
Section titled “Memory”The conversation so far, kept by a Chat Memory node so an agent can follow up on earlier messages. See Memory and context.
One step in a workflow, such as Get All Text, IF or Summarization. See Nodes and the node reference.
Output parser
Section titled “Output parser”A node you plug into an AI node so its answer comes out as fields or a list instead of free text. See Prompting and outputs.
Project
Section titled “Project”A folder for chats in the assistant. Every conversation in a project shares its instructions, memory and files. Projects group chats, not workflows. See Projects.
RAG (retrieval-augmented generation)
Section titled “RAG (retrieval-augmented generation)”Searching your documents first and giving the best passages to the model with the question, so the answer comes from your sources. See RAG.
One execution of a workflow, from the trigger to the last step. Past runs appear in the run history. See Workflow lifecycle and Run history.
Run variable
Section titled “Run variable”A value stored during a run by the Set Variable node and read later as {{ $run.name }}. See Mapping with expressions.
A connection point on an AI node where you plug in a model, memory, knowledge, tools or an output parser. See Models and where they run.
Template
Section titled “Template”A ready-made workflow you can copy and adapt, for example from the Marketplace. See Publishing.
A node an agent can choose to call, such as Web Search, Wikipedia or Browser. See Tool selection.
Trigger
Section titled “Trigger”The node that starts a workflow: a hotkey, a schedule, a page load, a chat message and so on. See Workflow lifecycle.
Vector store
Section titled “Vector store”Where embeddings are stored and searched. In AWFlow, that’s your knowledge bases, used through the Local Knowledge node. See Embeddings and vectors.
Workflow
Section titled “Workflow”Nodes connected together to do a job, from a trigger to the last step. See Workflow lifecycle.