--- title: Glossary description: A comprehensive glossary of key terms and concepts used in Langfuse documentation. --- # Glossary This glossary provides definitions for key terms and concepts used throughout the Langfuse documentation. Use the filters below to browse by category or search for specific terms. ## A ### Agent [#agent] Also known as: Observation Type An observation type that represents an AI agent workflow, including multi-step reasoning processes, tool orchestration, and autonomous decision-making. Used to track agent behavior and interactions. Categories: Observability Related: Observation, Tool, Agent Graph, Span [Learn more](/docs/observability/features/observation-types) ### Agent Graph [#agent-graph] A visual representation of complex AI agent workflows in Langfuse. Agent graphs help you understand and debug multi-step reasoning processes and agent interactions by displaying the flow of observations within a trace. Categories: Observability Related: Trace, Observation, Agent, Span [Learn more](/docs/observability/features/agent-graphs) ### Agentic Access [#agentic-access] Ways for AI agents to work with Langfuse data and workflows programmatically, via the Langfuse Agent Skill, CLI, or MCP server, across observability, evaluation, and prompt management. Categories: Platform Related: MCP Server, Public API, SDK, Langfuse CLI [Learn more](/docs/observability/features/agentic-access) ### AI Engineering Loop [#ai-engineering-loop] A lifecycle for continuously improving AI-powered systems by connecting production visibility with development workflows. It moves from tracing and monitoring real behavior to building datasets, running experiments, and evaluating changes before the cycle starts again. Categories: Observability, Evaluation Related: Trace, LLM-as-a-Judge, Dataset, Evaluator [Learn more](/academy/ai-engineering-loop) ### Annotation Queue [#annotation-queue] A manual evaluation method that allows domain experts to review and add scores and comments to traces, observations, or sessions. Useful for building ground truth, systematic labeling, and team collaboration. Categories: Evaluation Related: Score, Trace, Session, Online Evaluation [Learn more](/docs/evaluation/evaluation-methods/annotation-queues) ### API Key [#api-key] Credentials used to authenticate with the Langfuse API and SDKs. API keys consist of a public key and secret key and are associated with a specific project. They are managed in project settings. Categories: Platform, API Related: Project, Public API, SDK [Learn more](/docs/administration/rbac) ## B ### Billable Unit [#billable-unit] The unit of measurement for Langfuse Cloud pricing. Units are the sum of traces, observations, and scores ingested per billing period. Categories: Platform Related: Trace, Observation, Score [Learn more](/docs/administration/billable-units) ## C ### Chain [#chain] Also known as: Observation Type An observation type that represents a link between different application steps, such as passing context from a retriever to an LLM call. Categories: Observability Related: Observation, Span, Retriever, Generation [Learn more](/docs/observability/features/observation-types) ### Chat Prompt [#chat-prompt] Also known as: Message Prompt A prompt type that consists of an array of messages with specific roles (system, user, assistant). Useful for managing complete conversation structures and chat history. Categories: Prompts Related: Text Prompt, Prompt Management, Prompt Variables [Learn more](/docs/prompt-management/data-model#text-vs-chat-prompts) ### Custom Dashboards [#custom-dashboards] Flexible, self-service analytics dashboards that allow you to visualize and monitor metrics from your LLM application. Dashboards support multiple chart types, filtering, and multi-level aggregations. Categories: Observability Related: Score, Token, Pulse [Learn more](/docs/metrics/features/custom-dashboards) ## D ### Dataset [#dataset] A collection of test cases (dataset items) used to test and benchmark LLM applications. Datasets contain inputs and optionally expected outputs for systematic testing. Categories: Evaluation Related: Dataset Item, Dataset Experiment, Offline Evaluation [Learn more](/docs/evaluation/experiments/datasets) ### Dataset Experiment [#dataset-experiment] Also known as: Dataset Run, Experiment Run Also known as a Dataset Run. The execution of a dataset through your LLM application, producing outputs that can be evaluated. Links dataset items to their corresponding traces. Categories: Evaluation Related: Dataset, Dataset Item, Task, Score [Learn more](/docs/evaluation/experiments/data-model#datasetrun-experiment-run) ### Dataset Item [#dataset-item] An individual test case within a dataset. Each item contains an input (the scenario to test) and optionally an expected output. Categories: Evaluation Related: Dataset, Dataset Experiment, Task [Learn more](/docs/evaluation/experiments/data-model#datasetitem-object) ## E ### Embedding [#embedding] Also known as: Observation Type An observation type that represents a call to an LLM to generate embeddings. Can include model information, token usage, and costs. Categories: Observability Related: Observation, Generation, Retriever, Token [Learn more](/docs/observability/features/observation-types) ### Environment [#environment] A way to organize traces, observations, and scores from different deployment contexts (e.g., production, staging, development). Helps keep data separate while using the same project. Categories: Observability Related: Project, Trace, Tags [Learn more](/docs/observability/features/environments) ### Evaluation Method [#evaluation-method] A function that scores traces, observations, sessions, or dataset runs. Methods include LLM-as-a-Judge for subjective assessments, Annotation Queues for human review, Scores via UI for spot checks, and Scores via API/SDK for programmatic evaluation. Categories: Evaluation Related: Score, LLM-as-a-Judge, Annotation Queue, Dataset Experiment [Learn more](/docs/evaluation/core-concepts#evaluation-methods) ### Evaluator [#evaluator] Also known as: Scorer, Grader, Judge An observation type that represents functions assessing the relevance, correctness, or helpfulness of LLM outputs. Also refers to the function that scores experiment results. Categories: Observability Related: Score, LLM-as-a-Judge, Observation, Evaluation Method [Learn more](/docs/observability/features/observation-types) ### Event [#event] Also known as: Observation Type A basic observation type used to track discrete events in a trace. Events are the building blocks of tracing. Categories: Observability Related: Observation, Span, Trace [Learn more](/docs/observability/features/observation-types) ## F ### Filter Search Bar [#filter-search-bar] A single-line query bar for filtering and searching the Observations and Traces tables by typing field:value expressions instead of assembling filters in the sidebar. Supports operators, wildcards, negation, and an Ask AI button that drafts filters from a plain-language description. Categories: Observability Related: Trace, Observation, Tags, Environment [Learn more](/docs/observability/features/filter-search-bar) ### Flush [#flush] The process of sending buffered trace data to the Langfuse server. Important for short-lived applications to ensure no data is lost when the process terminates. Categories: SDK Related: SDK, Trace, Instrumentation [Learn more](/docs/observability/sdk/instrumentation#client-lifecycle--flushing) ## G ### Generation [#generation] Also known as: Observation Type An observation type that logs outputs from AI models including prompts, completions, token usage, and costs. The most common observation type for LLM calls. Categories: Observability Related: Observation, Token, Span [Learn more](/docs/observability/features/observation-types) ### Guardrail [#guardrail] Also known as: Observation Type An observation type that represents a component protecting against malicious content, jailbreaks, or other security risks. Categories: Observability Related: Observation, Trace, Agent [Learn more](/docs/observability/features/observation-types) ## I ### Instance Switcher [#instance-switcher] A self-hosted Enterprise Edition feature that lists your Langfuse deployments (e.g. development, staging, production) in the sidebar user menu so you can switch between them without keeping URLs elsewhere. Mirrors the region switcher on Langfuse Cloud. Categories: Platform Related: Organization, Project [Learn more](/self-hosting/administration/ui-customization#instance-switcher) ### Instrumentation [#instrumentation] The process of adding code to record application behavior. Langfuse provides context managers, observe wrappers, and manual observation methods for instrumenting your application. Categories: SDK, Observability Related: SDK, Trace, Observation, Flush [Learn more](/docs/observability/sdk/instrumentation) ## L ### Langfuse Assistant [#langfuse-assistant] An in-product AI assistant, available on Langfuse Cloud, for exploring project data and Langfuse workflows in plain language. It queries traces, observations, sessions, and metrics through the Langfuse MCP server, searches documentation, and proposes navigation actions that you confirm. Categories: Platform Related: MCP Server, Trace, Observation, Session [Learn more](/docs/langfuse-assistant) ### Langfuse CLI [#langfuse-cli] A command-line tool that wraps the Langfuse Public API, letting you manage prompts, evaluators, datasets, and other resources from the terminal or from AI coding agents. Supports pinning or auto-detecting a server's API version, so it works against self-hosted deployments running older releases. Categories: API, Platform Related: Public API, MCP Server, Agentic Access, SDK [Learn more](/docs/api-and-data-platform/features/cli) ### LLM Connection [#llm-connection] An API key configuration that allows Langfuse to call LLM models in the Playground or for LLM-as-a-Judge evaluations. Supports providers like OpenAI, Anthropic, and Google. Categories: Platform Related: Playground, LLM-as-a-Judge [Learn more](/docs/administration/llm-connection) ### LLM-as-a-Judge [#llm-as-a-judge] An evaluation method that uses an LLM to score the output of your application based on custom criteria. Provides scalable, repeatable evaluations with chain-of-thought reasoning. Categories: Evaluation Related: Score, Evaluator, Online Evaluation, Offline Evaluation [Learn more](/docs/evaluation/evaluation-methods/llm-as-a-judge) ### Log View [#log-view] Shows all observations concatenated. Great for quickly scanning through them. Categories: Observability Related: Agent Graph [Learn more](/docs/observability/overview) ## M ### MCP Server [#mcp-server] A Model Context Protocol server that enables AI-powered tools to interact with Langfuse data. Used for advanced integrations and AI-assisted workflows. Categories: Platform Related: Public API, SDK, Langfuse CLI [Learn more](/docs/api-and-data-platform/features/mcp-server) ### Metrics API [#metrics-api] An API endpoint for retrieving customized analytics from Langfuse data. Allows specifying dimensions, metrics, filters, and time granularity to build custom reports and dashboards for LLM applications. Categories: API Related: Custom Dashboards, Public API, Token [Learn more](/docs/metrics/features/metrics-api) ### Model Definition [#model-definition] A configuration that stores pricing information for an LLM model. Model definitions specify the cost per input and output token, enabling Langfuse to automatically calculate the price of generations based on token usage. Categories: Observability Related: Token, Generation, Custom Dashboards [Learn more](/docs/observability/features/token-and-cost-tracking) ## O ### Observation [#observation] An individual step within a trace. Observations can be of different types (span, generation, event, tool, etc.) and can be nested to represent hierarchical workflows. Categories: Observability Related: Trace, Span, Generation, Event [Learn more](/docs/observability/data-model) ### Offline Evaluation [#offline-evaluation] Testing your application against a fixed dataset before deployment. Used to validate changes and catch regressions during development. Categories: Evaluation Related: Dataset, Online Evaluation, Score, Dataset Experiment [Learn more](/docs/evaluation/core-concepts#the-evaluation-loop) ### Online Evaluation [#online-evaluation] Scoring live production traces to catch issues in real traffic. Helps identify edge cases and monitor application quality in production. Categories: Evaluation Related: Trace, Score, Offline Evaluation, LLM-as-a-Judge [Learn more](/docs/evaluation/get-started/online) ### OpenTelemetry [#opentelemetry] Also known as: OTel An open standard for collecting telemetry data from applications. Langfuse is built on OpenTelemetry, enabling interoperability and reducing vendor lock-in. Categories: Observability, SDK Related: Trace, Span, Instrumentation [Learn more](/integrations/native/opentelemetry) ### Organization [#organization] A top-level entity in Langfuse that contains projects. Organizations manage billing, team members, and SSO configuration. Categories: Platform Related: Project, RBAC [Learn more](/docs/administration/rbac) ## P ### Playground [#playground] The LLM Playground where you can test, iterate, and compare different prompts and models directly in Langfuse without writing code. Categories: Prompts Related: Chat Prompt, Text Prompt, LLM Connection [Learn more](/docs/prompt-management/features/playground) ### Project [#project] A container that groups all Langfuse data within an organization. Projects enable fine-grained role-based access control and separate data for different applications. Categories: Platform Related: Organization, RBAC, API Key, Environment [Learn more](/docs/administration/rbac) ### Prompt Label [#prompt-label] A label that can be assigned to a prompt version. Used to mark prompt versions as production or staging to fetch them via the SDK or API. Categories: Prompts Related: Prompt Management, Protected Prompt Label [Learn more](/docs/prompt-management/features/prompt-version-control) ### Prompt Management [#prompt-management] A systematic approach to storing, versioning, and retrieving prompts for LLM applications. Decouples prompt updates from code deployment. Categories: Prompts Related: Chat Prompt, Text Prompt, Prompt Variables, Playground [Learn more](/docs/prompt-management/overview) ### Prompt Variables [#prompt-variables] Placeholders in prompts that are dynamically filled at runtime. Allow creating reusable prompt templates with customizable content. Categories: Prompts Related: Prompt Management, Chat Prompt, Text Prompt [Learn more](/docs/prompt-management/features/variables) ### Protected Prompt Label [#protected-prompt-label] Restricts the ability to modify certain prompt labels (e.g. production) from being added to new prompt versions to admins and owners. This prevents accidental or unauthorized changes to production prompts. Categories: Prompts Related: Prompt Management, Environment [Learn more](/docs/prompt-management/features/prompt-version-control) ### Public API [#public-api] The REST API that provides access to all Langfuse data and features. Used for custom integrations, workflows, and programmatic access. Categories: API Related: SDK, API Key, MCP Server, Langfuse CLI [Learn more](/docs/api-and-data-platform/features/public-api) ### Pulse [#pulse] A compact outlier-chart strip above the Observations table that surfaces count, cost, and latency spikes over time. Clicking or dragging a spike narrows the table to that time window. Categories: Observability Related: Observation, Custom Dashboards [Learn more](/docs/observability/features/pulse) ## R ### RBAC [#rbac] Also known as: Role-Based Access Control Role-Based Access Control that manages permissions within Langfuse. Roles include Owner, Admin, Member, Viewer, and None, each with specific scopes. Categories: Platform Related: Organization, Project [Learn more](/docs/administration/rbac) ### Remote Experiment [#remote-experiment] A webhook-based trigger that allows running SDK experiments from the Langfuse UI. Configure a webhook URL and default config, then trigger experiments that fetch the dataset, run your application, and ingest scores back into Langfuse. Categories: Evaluation Related: Dataset, Dataset Experiment, Score [Learn more](/docs/evaluation/experiments/experiments-via-sdk#optional-trigger-sdk-experiment-from-ui) ### Retriever [#retriever] Also known as: Observation Type An observation type that represents data retrieval steps, such as calls to vector stores or databases in RAG applications. Categories: Observability Related: Observation, Chain, Embedding [Learn more](/docs/observability/features/observation-types) ### Rule [#rule] A configuration that selects incoming observations using filters and a sampling rate, then triggers one or more evaluators to score them. Categories: Evaluation Related: Evaluator, Observation, Online Evaluation, Score [Learn more](/docs/evaluation/core-concepts#evaluators-and-rules) ## S ### Score [#score] The output of an annotation or automated evaluation. Scores can be numeric, categorical, boolean, or text and are assigned to traces, observations, sessions, or dataset runs. Categories: Evaluation Related: Score Config, Evaluator, LLM-as-a-Judge, Annotation Queue [Learn more](/docs/evaluation/scores/data-model#scores) ### Score Config [#score-config] A configuration defining how a score is calculated and interpreted. Includes data type, value constraints, and categories for standardized scoring. Categories: Evaluation Related: Score, LLM-as-a-Judge [Learn more](/docs/evaluation/scores/data-model#score-config) ### SDK [#sdk] Also known as: Software Development Kit Software Development Kit. Langfuse provides native SDKs for Python and JavaScript/TypeScript that handle tracing, prompt management, and API access. Categories: SDK Related: Instrumentation, Flush, Public API [Learn more](/docs/observability/sdk/overview) ### Session [#session] A way to group related traces that are part of the same user interaction. Commonly used for multi-turn conversations or chat threads. Categories: Observability Related: Trace, User Tracking [Learn more](/docs/observability/features/sessions) ### Span [#span] Also known as: Observation Type An observation type that represents the duration of a unit of work in a trace. The default observation type for most operations. Categories: Observability Related: Observation, Trace, Generation, OpenTelemetry [Learn more](/docs/observability/features/observation-types) ## T ### Tags [#tags] Flexible labels that categorize and filter traces and observations. Useful for organizing by feature, API endpoint, workflow, or other criteria. Categories: Observability Related: Trace, Environment [Learn more](/docs/observability/features/tags) ### Task [#task] A function definition that processes dataset items during an experiment. The task represents the application code you want to test. Categories: Evaluation Related: Dataset, Dataset Item, Dataset Experiment [Learn more](/docs/evaluation/experiments/data-model#task) ### Text Prompt [#text-prompt] Also known as: String Prompt A prompt type that consists of a single string. Ideal for simple use cases or when you only need a system message. Categories: Prompts Related: Chat Prompt, Prompt Management, Prompt Variables [Learn more](/docs/prompt-management/data-model#text-vs-chat-prompts) ### Token [#token] The basic unit of text that LLMs process. Tokens can be words, parts of words, or characters depending on the model's tokenizer. Token counts determine API costs and context window limits. Langfuse tracks input and output tokens for cost monitoring and optimization. Categories: Observability Related: Generation, Custom Dashboards [Learn more](/docs/observability/features/token-and-cost-tracking) ### Tool [#tool] Also known as: Observation Type An observation type that represents a tool call in your application, such as calling a weather API or executing a database query. Categories: Observability Related: Observation, Agent, Span, Agent Graph [Learn more](/docs/observability/features/observation-types) ### Trace [#trace] A single request or operation in your LLM application. Traces contain the overall input, output, and metadata, along with nested observations that capture each step. Categories: Observability Related: Observation, Session, Span, Generation [Learn more](/docs/observability/data-model) ### Tracing [#tracing] The process of capturing structured logs of every request in your LLM application. Includes prompts, responses, token usage, latency, and any intermediate steps. Categories: Observability Related: Trace, Instrumentation, SDK [Learn more](/docs/observability/overview) ## U ### User Tracking [#user-tracking] The ability to associate traces with users via a userId. Enables per-user analytics, cost tracking, and filtering. Categories: Observability Related: Trace, Session [Learn more](/docs/observability/features/users) --- ## Agent Instructions This page is part of the [Langfuse](https://langfuse.com) documentation, published as plain Markdown for AI agents. Every page is available as Markdown by appending `.md` to its URL, or by sending an `Accept: text/markdown` header. This page: `https://langfuse.com/docs/glossary.md`. ### Querying these docs If the answer is not on this page, query the documentation instead of guessing: - **Semantic search** across all Langfuse docs, returning an answer with the relevant pages and excerpts. Ask a specific, self-contained question: ```bash curl -sG "https://langfuse.com/api/search-docs" --data-urlencode "query=How do I trace a LangGraph agent?" ``` - **Index of every page**: , with per-section indexes [llms-docs.txt](https://langfuse.com/llms-docs.txt), [llms-integrations.txt](https://langfuse.com/llms-integrations.txt), and [llms-self-hosting.txt](https://langfuse.com/llms-self-hosting.txt). ### Before writing Langfuse code - **Install the [Langfuse Agent Skill](https://langfuse.com/docs/api-and-data-platform/features/agent-skill).** It encodes Langfuse's own best practices for instrumentation, prompt management, and evaluation, and materially improves results. - **Read [What does a good trace look like?](https://langfuse.com/docs/observability/best-practices.md)** before instrumenting an application. - **Verify endpoints, parameters, and response fields** against the [API reference](https://api.reference.langfuse.com) instead of inferring them from code examples. - **Use the [Langfuse CLI](https://langfuse.com/docs/api-and-data-platform/features/cli)** (`npx langfuse-cli api `) to read or write traces, prompts, datasets, and scores from the terminal. Found an error in these docs? Please open an issue at .