Behind-the-scenes research, prompt development, evaluation, and experimentation for the Footprints project.
Footprints helps people reflect on ChatGPT conversation history by surfacing sustained projects, recurring themes, persistent questions, shifts in focus, and unfinished ideas. The goal is reflection, not productivity scoring, behavioural optimisation or diagnosis.
The central question is:
Looking back across a period of conversations, what emerged?
This repository is the source of truth for Footprints methodology.
It holds:
- ChatGPT export structure research;
- prompt experiments and iteration history;
- evaluation criteria;
- sample outputs;
- report design decisions;
- known limitations;
- open research notes.
The Custom GPT and Codex Skill repositories consume the outputs of this work.
Reports should help users recognise patterns in their own conversations. They should not score, rank, judge, diagnose, or profile people.
Claims should be grounded in the user's conversation history. If a conclusion cannot be reasonably supported, it should be softened or removed.
Footprints focuses on long-term movement, recurring themes, and meaningful returns rather than isolated messages.
Frequency matters, but it is not the same as importance. A brief thread may be meaningful if it is specific, connected to action, or returns later in another form.
A strong report should:
- identify major projects accurately;
- detect recurring themes and persistent questions;
- recognise long-term interests and shifts in focus;
- avoid excessive repetition and recency bias;
- distinguish evidence from interpretation;
- avoid unsupported claims;
- feel useful and recognisable to the user.
The ideal reaction is:
"Yes, that feels like the path I walked."
Current and future work includes:
- reflective report structure;
- idea archaeology;
- timeline and season detection;
- custom awards and highlights;
- export parsing and preprocessing;
- visual, presentation, and interactive formats.
- footprints-custom-gpt: public Custom GPT experience.
- footprints-codex-skill: open-source Codex Skill implementation for contributors and experimentation.
Experimental. The methodology, prompts, structure, and outputs are expected to evolve as more exports are tested and evaluated.
People leave footprints everywhere: in notebooks, projects, conversations, and unfinished ideas.
Footprints exists to help make those paths visible again.
MIT