rseng-project-kickoff

# Project kickoff: ask well, then decide well The first hour of a project fixes decisions that cost months to change later. A short interview beats both extremes: asking nothing (the agent guesses the user's constraints) and asking everything (the user abandons the questionnaire). Ask only what changes decisions, derive the rest, and write down which is which. ## The interview Eight questions cover what defaults cannot infer. Ask them in the user's language, batched sensibly, skipping any the context already answers: 1. Goal and audience: what research question, who runs this software - just you, your group, a community? (Sets the tier - rseng-quality-framework - which calibrates everything else.) 2. Stack: a language/ecosystem preference, or should the agent choose per the state of the art for the domain (rseng-language-guides, verified online when unfamiliar)? 3. Data: what data, how big, and is any of it personal or sensitive? (Routes rseng-data-management and rseng-regulatory-compliance before the first file lands in git.) 4. Openness: license preference and open-science intent - open from day one or open at publication (rseng-licensing, rseng-open-science-practices)? 5. Collaboration: solo or team, internal or external contributors (rseng-community-governance's scale)? 6. Compute: laptop, cluster, GPUs (rseng-hpc-computing, rseng-gpu-computing shape structure early)? 7. Deadlines: the research calendar dates that planning works backwa...