FSharp.Stats is primarily maintained through agentic development under human guidance. Maintainers discuss proposed work and then direct coding agents to make the complete change, including implementation, tests, documentation, benchmarks, and other affected files.
We generally prefer contributions as GitHub issues rather than pull requests. Search for an existing report first. Issues may include proposed changes, patches, reference implementations, or links to forks or branches. Maintainers may refine the scope and assign the issue to an agent to implement and validate the complete change.
Repo Assist is an automated AI assistant that runs regularly in this repository. It may triage or respond to issues, investigate bugs, suggest improvements, and attempt implementations as draft pull requests. Its work is identified as automated and remains subject to human review; Repo Assist does not merge pull requests or make final maintenance decisions.
Maintainers can invoke Repo Assist with /repo-assist <instructions> for a specific agentic task,
such as investigating an issue, preparing a fix, adding tests, or updating documentation.
Every pull request must have a matching issue that has been discussed with the maintainers. Link the
pull request to that issue, keep it focused, and target the developer branch. Maintainers may close
a pull request and use the issue as the basis for an agent-produced implementation instead; the
submitted analysis and code remain valuable inputs to that work.
If a pull request is the agreed approach, follow AGENTS.md. Statistical and machine-learning changes need tests and a canonical reference implementation or an explicit alternative validation method. Before submission, run:
./build.sh RunTests