Colophon

Records AI writing interactions as a TWFF-compatible process log in Google Docs.

Colophon documents *how* a piece of writing was made — not to police it, but to make the process visible and verifiable. Working inside Google Docs, Colophon records your writing session as a structured, tamper-evident log: what you typed, when you used AI assistance, and what you decided to do with each AI suggestion (accept, modify, or reject). You stay in control — the log is yours, it stays on your machine, and you choose if and when to export and share it. The result is a portable .twff file: an open, JSON-based "process log" that shows the story behind a finished document. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ WHAT IT CAPTURES ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ • Native Gemini suggestions in Google Docs — detects when "Help me write" proposes text and records whether you inserted or dismissed it. • Edits and typing — aggregated into edit events, with typing-speed signals that help distinguish authored text from pasted or generated blocks. • Pastes — captured separately, so external content is never mistaken for typing. • A live timeline — a side panel shows your writing activity as it happens. • Integrity — every event is chained with a SHA-256 hash, so a log can be verified and any tampering detected. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ WHO IT'S FOR ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ • Students and educators who want a transparent, good-faith record of original work. • Writers and researchers who value provenance and reproducibility. • Anyone who wants to show the effort behind their writing in an age of AI. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ PRIVACY ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ • Local-first: your process log is stored on your device. • Nothing is uploaded automatically — exporting the .twff file is a deliberate action you take. • Colophon only runs on Google Docs document pages. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ OPEN STANDARD ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ Colophon writes to TWFF, an open format for writing-process logs. Learn more and inspect the specification: https://github.com/Functional-Intelligence-Research-Lab/twff Colophon is developed by the Functional Intelligence Research Lab.