Glossary

Terms from AnyLLM and from Jira and Confluence — definitions explained as far as you need them to use the app — followed by a comparison of the English and German interface labels (button names, menu entries, messages). Each term links to the page that covers the topic in depth.

Terms A–Z

Agent An assistant with fixed instructions for a particular purpose — for example a "Sprint reporter" that always answers briefly and in table form. Three kinds: provided by the administration (Predefined (admin)), your own (My agents) and those shared by colleagues (Shared by colleagues). Your agents are the same in Jira and Confluence. → Creating your own agents and skills

API key The password towards an AI service. It is stored encrypted at Atlassian, is never visible in plain text and can only be replaced or deleted. → Setting up LLM connections

App page (shell mode) The full chat under Apps → AnyLLM Chat in Jira or Confluence. It has no header row: the sidebar carries the AnyLLM brand, the sections Chat, Agents and Skills, the History list and, at the bottom, Settings (administrators only), Switch to Confluence / Switch to Jira and Open as full page. The fullscreen overlay and the standalone full page look the same. → Getting started

Archive Saving a copy of a conversation as a page under "AnyLLM Chats" in your personal Confluence space (Archive to Confluence). The conversation stays in your history; the page is as visible as your personal space. → Managing chat history

Byline assistant The entry AnyLLM Chat below a Confluence page title. It opens the chat in a tall, narrow dialog that knows the page. → AnyLLM in Confluence

Chat bar (page banner) The ask-anything bar at the top of Confluence pages with the field Ask AnyLLM anything — e.g. about this page … and the buttons Chat, Continue conversation and Dismiss. Chat opens the fullscreen overlay with your question. → AnyLLM in Confluence

Chip A small blue button next to the input field indicating an active special state — for example Temporary ✕ or the memory chip. → Chatting with the AI

Composer The input area at the bottom of the chat: text field, ＋ menu, agent and model selection, chips and the send button. → Getting started

Connection An AI service set up by the administration: address (endpoint), API key and list of models. Up to five per installation — shared by Jira and Confluence. → Setting up LLM connections

Context (page context) The content of the open page or issue (plus the latest comments), which is attached to a question while Include page context is switched on in the ＋ menu. It is re-read with your own permissions for every message. → Prompts, skills, memory and model parameters (＋ menu)

Conversation One entry in your chat history (History in the sidebar). Titled automatically, kept up to 50 per person, identical in Jira and Confluence. → Managing chat history

CQL The search language of Confluence, the counterpart to JQL. The AI uses it when it searches for pages. → Letting the AI search, read and create

Default model The model that is preselected for users as long as they have not chosen one themselves. Set by the administration under General. → General settings

Egress → see Endpoint approval

Endpoint The internet address of an AI or search service to which AnyLLM sends requests. Always an https address; for OpenAI-compatible services including /v1. → Setting up LLM connections

Endpoint approval The explicit permission for the app to talk to a particular address; without it the request is blocked inside Atlassian. The list of approved domains is kept once for the whole installation and looks the same in Jira and Confluence, but the administrator's consent in the Atlassian dialog counts per product — grant it once in Jira and once in Confluence. → Managing endpoint approvals

Fork A copy of a conversation in which you continue working without changing the original (Fork — continue in a copy). The copy is prefixed with "⑂". → Managing chat history

Fullscreen overlay The chat as a full-screen window titled AnyLLM on top of the issue or page you are on, opened with ⤢ Fullscreen from panels and dialogs or with Chat in the Confluence chat bar. It looks like the app page and is closed with its own ✕. → AnyLLM in Jira

Gadget A building block on a Jira dashboard. The AnyLLM gadget (listed as AnyLLM Chat) answers a saved question and shows the answer to everyone who can see the dashboard. → Dashboard gadget

Instructions (system prompt) The text that tells an agent how to behave. It is placed before every request but is not visible in the chat; other people's instructions cannot be viewed. → Creating your own agents and skills

Issue sidebar panel The entry AnyLLM in the right-hand context area of a Jira issue. It opens a compact chat of fixed height that knows the issue. → AnyLLM in Jira

JQL The search language of Jira, which allows issues to be filtered precisely. The AI uses it when it searches for issues; the tool step row shows the query. → Letting the AI search, read and create

LLM (large language model) The language model behind the answers — in this manual usually simply "the model" or "the AI". Which service is used is decided by the administration's connections. → Setting up LLM connections

Macro A building block that can be inserted into a Confluence page. The AnyLLM macro embeds the full chat directly into the page. → AnyLLM in Confluence

Memory Personal notes that are sent with every request — for example your role or your preferred answer style. Visible only to you, at most 4,000 characters, the same in Jira and Confluence; switch it on or off with Memory active — sent with every request. → Prompts, skills, memory and model parameters (＋ menu)

Model The concrete flavour of the AI, e.g. gpt-4o. Models differ in speed, cost and quality; which ones are available is determined by the administration. → Chatting with the AI

Model parameters Temperature (factual ↔ creative) and Max response tokens (answer length), set in the ＋ menu under ⚙ Model parameters …. → Prompts, skills, memory and model parameters (＋ menu)

Module A single surface of the app — issue panel, macro, chat bar and so on. The administration can switch each of the 14 modules on and off under Modules; the toggles apply to Jira and Confluence together. → Modules

Placeholder A gap in a template or skill, written in double curly braces, e.g. {{Project}}. When the template is used, the value is asked for. → Prompts, skills, memory and model parameters (＋ menu)

Product compatibility AnyLLM is one app installation serving Jira and Confluence on the same site: Jira is the required product, Confluence is connected afterwards. Both share one configuration and one data store; only the endpoint-approval consent is granted per product. → Administration & configuration

Prompt / Template The text you send to the AI. A template (also "prompt template") is a prepared prompt provided by the administration under Prompts & Skills; it is inserted into the message box via the ＋ menu. → Prompts, skills, memory and model parameters (＋ menu)

Proposal card → see Suggestion card

Provider type The way in which AnyLLM technically talks to an AI service: OpenAI-compatible, self-hosted, Azure OpenAI or Anthropic. → Setting up LLM connections

Quick actions One-click actions in Confluence: on selected text (Explain, Summarize, Translate, Improve writing) and on the whole page via the ••• menu (Summarize page, Extract action items, Translate page). Every result is also saved as a new conversation in your history. → AnyLLM in Confluence

Rate limit The upper limit for the number of requests per person and hour, set by the administration. One counter per person covers Jira and Confluence together; every send counts. → General settings

Site knowledge A text about your own organisation maintained by the administration (project keys, terminology), which is sent to the AI with every request in both products. → General settings

Skill A template with one-click execution: click it in the ＋ menu, fill in placeholders if there are any, and the question is sent immediately — optionally always with a particular agent. You can create your own skills. → Creating your own agents and skills

Standalone full page The chat on its own browser page without the Jira or Confluence frame. Reached only via Open as full page at the bottom of the sidebar; administrators enable it under Modules → Across products. → AnyLLM in Jira

Streaming The answer appears word by word as it is produced, instead of only at the end. While the model works you see Model is thinking …. → Chatting with the AI

Suggestion card The card with which the AI proposes a writing action. Four types: Create Jira issue, Update Jira issue, Add Jira comment, Create Confluence page. Only ✓ Execute triggers it; Dismiss removes the card. → Letting the AI search, read and create

Temperature Controls how freely the AI phrases things: low = predictable and precise, high = more creative. → Prompts, skills, memory and model parameters (＋ menu)

Temporary chat A mode in which the conversation is not saved (Temporary chat (not saved) in the ＋ menu, chip Temporary ✕). Export it before you leave. → Prompts, skills, memory and model parameters (＋ menu)

Token The billing unit of language models — roughly a part of a word. The usage statistics count input and output tokens. → Usage statistics, import/export and backup

Tool call (tool step) A lookup the AI performs before answering — a Jira or Confluence search, reading an issue or page, a web search or an image generation. Each step is shown as a row with an English label such as "Jira search: …"; errors go back to the AI, which may paraphrase them. → Letting the AI search, read and create

Tool level Determines per agent how much access to Jira and Confluence it may have (Jira/Confluence tools: Like global policy, Off, Read only, Read + write). It can only narrow the administration's AI access setting, never widen it. → Creating your own agents and skills

English and German interface labels

As shipped, the interface is in English; the administration switches it under General → Interface language (General settings). Menu titles in Jira and Confluence (AnyLLM Chat) follow your Atlassian language, and server messages are always English — except the endpoint-approval message listed under Chat. In the tables, {p} stands for the product name (Jira or Confluence) and {n}, {count}, {limit}, {list}, {date} and {format} for values filled in at runtime; labels of one control group share a row, separated by " · ".

Chat

Settings (administration)

Confluence quick actions and chat bar

Dashboard gadget

Related topics

• AnyLLM — User Manual — overview and index of all pages
• Getting started — interface tour: sidebar, header and composer
• General settings — switching the interface language
• Modules — the 14 surfaces and their toggles
• Troubleshooting — messages quoted verbatim, with causes and solutions
• FAQ — short answers for users and administrators