Contributing

June 7, 2026 ยท View on GitHub

As an open-source project we welcome contributions in all areas: to the DSP library, to Tiliqua's platform code, to the examples, to the Rust firmware - whatever piques your interest. As long as you're happy with your contribution falling under the same license as the project.

In short, contributions to Tiliqua can happen in 3 areas:

  • Out-of-tree: you fork the repo, change some stuff, share the repo or bitstreams with people, keeping things open source but never merging your changes back.
    • You may file a pull request against tiliqua-webflash with a link to your source, and I would be happy to review the bitstream and test it, making it available to all users.
    • Here the review hurdle is 'it is interesting, it works and won't damage Tiliqua or user hardware' - the contribution rules below do not apply to out-of-tree contributions, beyond the code being reviewable and proven to work.
  • In-tree toplevels: you make some changes to the included top-level bitstreams. Here the contribution rules below DO apply.
  • In-tree shared libraries / hardware: you make changes to infrastructure shared between all Tiliqua projects. Here, the contribution rules below DO apply.

Use of LLMs / AI in contributions

This applies for any PRs against this repository.

As part of NGI0 (see Funding below), please become familiar with NLnet's AI policy. Below are some additional guidelines specific to this project.

Put simply: any PR which looks completely AI-generated, especially where it is doubtful the author understands what they have changed / added, will be rejected.

This project does not accept AI-assisted contributions in the following areas:

  • All code shared amongst top-levels - such as DSP library code and platform drivers.
  • Any top-level bitstreams used in tutorials and intented to be pedagogical (dsp/top.py).
  • All documentation text.
  • All hardware design and documentation.

This project may accept AI-assisted contributions with clear provenance, which are reviewable and clearly explainable in the following areas:

  • Any top-level bitstreams not covered above.
  • Test and simulation harnesses

Commit messages, PR titles and descriptions must always be hand-written. In addition to NLnet's AI policy, some points borrowed from betrusted-io/xous-core:

- When you submit a contribution, you represent that you are the author and that you are fully accountable for the entirety of the contribution.
- You are responsible for your contribution, including vouching for the quality, license compliance, and utility of your submission.
- Using AI assistance in your contribution does not relieve you of the responsibility to ensure that your contribution meets project standards; your contributions must be modular, reviewable, and clearly explainable.