oxo-call: An Intelligent Command-Line Assistant for Bioinformatics, Turning Natural Language into Professional Tool Calls in Seconds

oxo-call is a Rust-based intelligent command-line assistant for bioinformatics, whose core goal is to convert natural language task descriptions into accurate professional tool commands. It addresses the high barrier to using bioinformatics command-line tools through a document-first grounding strategy (providing full version-specific help text of the target tool before generating commands) and a curated skill enhancement strategy (with over 150 built-in skills covering 44 types of analysis). It also supports features like a DAG workflow engine to facilitate reproducible research.

Bioinformatics relies heavily on command-line tools, which have complex parameters, strict format requirements, and significant version differences. Beginners or cross-domain users face a steep learning curve, and parameter errors can easily lead to wasted computation or incorrect results. Among existing solutions, graphical platforms like Galaxy sacrifice flexibility and efficiency, while direct command-line use requires deep professional knowledge. Balancing flexibility and low barriers has long been a pain point.

oxo-call is optimized for bioinformatics needs: 1. Document-first grounding: Provides full version-specific help text of the target tool to the LLM before generating commands, reducing parameter hallucinations; 2. Curated skill enhancement: Over 150 built-in skills (covering 44 types of analysis) include domain expert concepts, common pitfalls, and working examples, compiled by experts to assist in generating correct commands and workflows.

The technical features of oxo-call include: 1. Single static binary file: No complex dependencies, run immediately after download, easy for cluster deployment; 2. DAG workflow engine: Connects multi-step analyses, automatically handles dependencies and data transfer; 3. Traceability: Records metadata such as tool versions, parameters, and inputs of commands to support reproducibility; 4. Privacy protection: Supports local LLM inference, no data sent to the cloud; 5. Extensibility: Customize or import skills via the MCP protocol.

The value of oxo-call: 1. Lowering barriers: New users can describe their needs in natural language and learn by observing generated commands; 2. Improving efficiency: Helps senior researchers quickly generate commands, saving time spent searching documentation; 3. Standardizing workflows: The skill system enforces standard processes to ensure team consistency; 4. Promoting sharing: The skill sharing mechanism makes it easy for teams to package workflows for collaborators or the community.

Current limitations: Mainly supports academic use; commercial use requires additional authorization; local LLM inference has certain hardware requirements. Future directions: Support more tools and databases; integrate visualization features; enhance error handling capabilities; develop domain-specific professional skill packages.

oxo-call does not replace researchers; instead, it frees them from tedious command details to focus on scientific problems. It is worth trying for bioinformatics researchers. Project address: https://traitome.github.io/oxo-call/; Paper link: http://arxiv.org/abs/2604.12387v1.