Xorq

Xorq is an engine for building portable, multi engine data pipelines, described in its documentation as an agentic execution system for AI. A developer writes expressions in Python, and Xorq defers their execution, caches intermediate results, tracks lineage and lets individual steps be routed to different backends, so the same pipeline can move between engines without being rewritten. A central idea is the catalog, a git based store used to publish, discover and reuse pipelines, which the docs call executable memory. Artifacts are content addressed, and pinned data artifacts are versioned. The Python API covers core operations, user defined functions, machine learning operations, caching, lineage, Flight operations and catalog operations, and a command line offers build, run, run cached, serve and catalog commands with 32 subcommands. Documentation includes a quickstart, how to guides for connecting to and mixing backends, tutorials on building a semantic catalog, and machine learning tutorials covering data splitting, training a first model and comparing model performance. A Claude Code plugin is documented in the getting started section, so a coding agent can work with Xorq pipelines. The tool installs with pip install xorq, and the source is on GitHub. The listing points to the documentation site, which publishes no pricing page or plan prices.