Feedbacker is being rebuilt as a workflow, governance, and assessment layer around AI for higher education.
The project helps educators operationalise assessment responsibly, consistently, and at scale. It does not seek to automate academic judgement. Educators remain responsible for marks and for reviewing, editing, and approving feedback before release.
Feedbacker's focus is shifting from generating feedback to providing dependable assessment infrastructure:
- rubric-led and repeatable workflows;
- consistency across markers, submissions, and cohorts;
- traceable assessment inputs, decisions, and outputs;
- moderation and calibration support;
- auditability and institutional quality assurance;
- explicit privacy, governance, and responsible-AI controls;
- institutional control over models, prompts, data, and deployment;
- efficient cohort-scale operation with educators firmly in control.
See the project definition and architecture principles for the current foundation.
The previous feedback-generation application has been retired from the active branch and preserved in legacy/v1-feedback-generator. The replacement is currently in its definition and architecture phase.
The holding page for feedbacker.education lives in site/ and is deployed to GitHub Pages.
Read AGENTS.md before making changes. Product and technical proposals should preserve educator control, traceability, privacy, accessibility, and responsible assessment practice.