Bringing Our "Walking Skeleton" to Life

This month, the development team has been working on something appropriately seasonal 🎃 a walking skeleton 🦴 version of our product. In software terms, this is a minimal end-to-end version of the system that allows us to test each stage of the process in a realistic way before adding more complexity.

Our walking skeleton demonstrates the full journey:

1. A researcher writes and tests their analysis code locally on synthetic data.
2. The researcher submits a request to run their code on the “real” data.
3. Staff receive the request, download the researcher’s code and instructions from GitHub, and run it in a secure, controlled environment.
4. Staff make the outputs available for the researcher to view.
5. The researcher requests permission to download and share the outputs.
6. Staff review the outputs to ensure they are fully anonymous (non-disclosive) and release them if appropriate.

To test this process, we took a phased approach:

• First, two members of the development team ran through the process internally, helping us refine the wording and order of steps in the playbook.
• Next, a member of staff and a developer (acting as a researcher) tested it, uncovering a few more points for improvement.
• Finally, we carried out a test with a staff member and an actual research user. This surfaced a significant issue related to permissions on the researcher’s laptop, as well as some smaller tweaks to make the playbook clearer.

Each round of testing has helped strengthen our walking skeleton - ensuring that when we start to build out the full system, the foundations are solid. It’s been a fitting project for October: a reminder that even the simplest skeleton needs care and attention before it can truly come to life. 👻