# Rseng Project Kickoff > Covers starting a new research software project with a short kickoff interview: asking the user the few questions whose answers actually change decisions (goal, software tier, stack, data sensitivity, openness, collaboration, compute, deadlines, and how much the agent may decide alone), then deriving tier-appropriate defaults, setting the project up, and making proactive decisions afterwards while keeping the user informed. Use PROACTIVELY when a new project is starting in an empty or fresh directory, and when the user asks to kick off, bootstrap or set up a new research software project or wants the agent to interview them about it. (Drafting SMPs is rseng-management-planning; steady-state operation after setup is rseng-project-tracking.) - Skill: `fdiblen/rseng-project-kickoff` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add fdiblen/rseng-project-kickoff` - Raw SKILL.md: https://api.skillmd.com/api/skills/fdiblen/rseng-project-kickoff/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - License: CC-BY-4.0 - Author: fdiblen (https://skillmd.com/u/fdiblen) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/fdiblen/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 backwards from (rseng-project-tracking). 8. The delegation contract - the question most kickoffs skip: which decisions may the agent make alone (tooling, structure, dependencies within policy) and which always come back to the user (scientific choices, anything on the ask-list they name, spending, publishing)? Record the answer; it governs the whole collaboration (rseng-pair-programming's boundary, made explicit per project). Offer sensible defaults with every question so "accept all" is a one-word answer for users who want speed. ## From answers to a running project Derive and execute, don't re-ask: tier-appropriate scaffolding (rseng-project-scaffolding), version control with guards (rseng-version-control-review, rseng-security's ignore rules before any secret exists), environment and lockfile per the chosen stack (rseng-reproducible-environments), LICENSE and citation files (rseng-licensing, rseng-citation-metadata), aidecl.yaml from the start (rseng-ai-declaration), a minimal tracker and the deadline milestones (rseng-project-tracking), and - when answers warrant - the data-sensitivity guards, the SMP/DMP skeletons (rseng-management-planning, rseng-data-management-plans) and a discovery pass over prior art (rseng-discovery). ## Deciding proactively, keeping the user informed After kickoff the delegation contract is the operating agreement: - Decisions inside the contract are MADE, not re-asked - and each consequential one is reported as a compact decision note at delivery: what was decided, the one-line why anchored in the user's answers ("pytest, because you chose Python and asked for team-friendly tooling"), and how to change it. Silence is not an option; neither is a question that was already answered. - Decisions outside the contract come back as a short question with a recommended default. - The kickoff answers and every later decision note land in the decision log (rseng-project-tracking), so the project's WHY survives the chat session. - Revisit the contract at milestones: trust earned can widen it; surprises can narrow it - either direction is a normal adjustment, not a failure (rseng-lessons-learned records the trigger). ## Working with this skill This skill is source-independent: it encodes kickoff-interview practice for AI-assisted research software projects. It orchestrates the pack's setup skills and hands steady-state operation to rseng-project-tracking. Learn more (verified): - https://opensource.guide/starting-a-project/ - starting an open source project - https://learn.scientific-python.org/development/ - Scientific Python development guide - https://book.the-turing-way.org - The Turing Way handbook - https://www.software.ac.uk/resource-hub - Software Sustainability Institute resources ## Related skills Check whether any of these applies before moving on: - rseng-data-management - data questions route here early - rseng-discovery - prior-art pass before building - rseng-management-planning - SMP skeleton after kickoff answers - rseng-project-scaffolding - executes the scaffolding step - rseng-project-tracking - hands over steady-state operation - rseng-quality-framework - tier classification calibrates all defaults