February 2026 Summaries

Langfuse's February 2026 update introduces a revamped observation-centric data model designed to enhance speed and scalability, resulting in a faster user interface and evaluation workflows. This shift focuses on observations as the primary data object, eliminating the need for deduplication operations and joins, which simplifies and accelerates database queries. The update also includes the official release of the Observations API v2 and Metrics API v2 out of beta, enabling faster querying and aggregation at scale. Users can experience significantly reduced chart loading times and improved responsiveness in browsing and filtering large projects. Additionally, observation-level evaluations now execute more swiftly as they bypass the need for querying ClickHouse for each evaluation. Langfuse provides a migration guide to help users unlock these performance improvements, depending on their specific needs, such as upgrading the SDK for real-time data or adopting the new APIs for quicker data retrieval at scale.

Langfuse utilized datasets, tracing, and the Claude Agent SDK to enhance an AI agent skill designed for accessing Langfuse's API, documentation, and observability practices. By treating skill evaluation like prompt evaluation, they stored user prompts in datasets and traced agent behaviors, iteratively improving the skill's quality. Initial challenges included the agent's frequent CLI errors, unnecessary retries, and incorrect usage of commands, which were addressed by enforcing mandatory parameters and adding proactive discovery steps. A restructuring of the skill's description initially led to its non-invocation, prompting a return to a more detailed explanation. Evaluating complex tasks like application instrumentation required using an LLM as a judge to verify the agent's modifications. Through detailed trace reviews and iterative adjustments, Langfuse identified areas for further enhancement, such as reducing CLI calls and refining auto-instrumentation in complex cases, with ongoing improvements and best practice documentation planned for the future.

The text discusses a method for improving prompts using an AI agent and the Langfuse skill, which involves analyzing trace feedback to enhance prompt quality iteratively. By annotating traces in Langfuse, users can provide feedback that the AI agent, demonstrated with the Claude model, analyzes to suggest prompt changes, rapidly progressing from a basic to a more effective prompt. This process involves examining traces, identifying errors or irrelevant responses, and categorizing them for better analysis. The example application used is a chatbot that searches GitHub discussions, illustrating how traces are recorded and evaluated. Claude assists in fetching scores, identifying gaps in prompts, and proposing improvements, with the iterative process continuing until most issues are resolved. The guide also highlights variations of this approach, such as using user feedback, annotation queues, or experiment results, and suggests creating a dataset for more structured evaluation to ensure prompt changes do not introduce regressions.

Langfuse has been enhanced with a suite of features aimed at making it an ideal platform for AI agents, focusing on integration and usability for developers. Key features include the Langfuse CLI, which provides command-line access to the entire REST API, and the introduction of Skills, which offer structured workflows to guide agents in utilizing Langfuse effectively. Documentation is accessible in markdown format, allowing agents to bypass unnecessary HTML clutter, thereby improving efficiency. A public RAG endpoint is available, enabling agents to query all indexed documentation and GitHub issues, while llms.txt offers a machine-readable index to quickly understand the documentation structure. Additionally, MCP servers facilitate access to Langfuse documentation and project data, with authentication protocols ensuring secure access. These improvements aim to streamline the interaction between AI agents and Langfuse, enhancing the overall agent experience and effectiveness in real-world applications.