Academic Integrity

What it does

• Stops writes to files matching data/raw/**.
• Stops writes to files matching **/raw-data/**.
• Stops writes to files matching **/raw_data/**.
• Tells every agent, when its session starts: "Never invent a citation. Cite only sources you have opened, give page or section numbers, and mark any reference you could not check."
• Tells every agent, when its session starts: "Treat raw data as read-only: work on copies, and never drop an observation or outlier without recording why."
• Tells every agent, when its session starts: "Keep a methods log of every analysis step, parameter and software version, so someone else could reproduce the result."

Peregrini chose this publisher. It has not checked each rule. A book cannot vary the Mandate, and no instruction in it permits anything the Mandate reserves to your own permission lines. Its blocks are added to yours: where two books block different things, both apply. Where its instructions conflict with your own, the agent says so and asks.

The exact lines

no-write data/raw/**
no-write **/raw-data/**
no-write **/raw_data/**
instruction Never invent a citation. Cite only sources you have opened, give page or section numbers, and mark any reference you could not check.
instruction Treat raw data as read-only: work on copies, and never drop an observation or outlier without recording why.
instruction Keep a methods log of every analysis step, parameter and software version, so someone else could reproduce the result.

Versions

1. Version 1 · 28 September 2026

   A starter book by Peregrini.

   Added: no-write data/raw/** · no-write **/raw-data/** · no-write **/raw_data/** · instruction Never invent a citation. Cite only sources you have opened, give page or section numbers, and mark any reference you could not check. · instruction Treat raw data as read-only: work on copies, and never drop an observation or outlier without recording why. · instruction Keep a methods log of every analysis step, parameter and software version, so someone else could reproduce the result.