spcaeo/skald is a self-hosted web analytics tool that analyzes server access logs. The latest release in the provided record is v2.14.1. It is written in Python and uses SQLite and Docker. It does not install JavaScript analytics scripts or cookies in the browser, and it separates regular visitors from AI assistants such as ChatGPT, Claude, and Perplexity, as well as from crawlers and bots. However, the provided record does not separately state what changed in v2.14.1.
The basis of the analysis is not signals sent from the browser but requests recorded by the server. The purpose of this project is therefore to also look at visits that JavaScript-based analytics tools miss. According to the README description, every figure is recounted by a separate, independent audit parser. It is a structure that verifies numbers along two paths, but the provided information contains no performance measurements comparing their accuracy. It is an approach that suits sites wanting to distinguish AI assistant traffic without placing tracking scripts in visitors' browsers.
The license information has to be read carefully. The repository's license field is marked as NOASSERTION, while the README summary states that it can be hosted directly for free under the Elastic License 2.0. Those two pieces of information alone are not enough to establish the usage terms. The log-based analytics capability is therefore confirmed, but the repository's actual license conditions should be examined separately before adoption.