Sprint 4: Onboarding refinement, knowledge quality & project setup (2026-07-01 to 2026-07-15)

[adoLime]

Sprint Goal

Members and PMs can shape the onboarding experience day-to-day (reorder, knowledge checks, freshness), Admins can set up projects and map users to them, and the chat becomes more precise (source / time filters, short summaries) — built on the AI onboarding & role foundation from Sprint 3.

In scope

A. Onboarding refinement (PM/Admin + User)

• PM/Admin reorders onboarding steps — drag-and-drop, audit, state-aware (Backend/Frontend) [Story]: PM/Admin reorders onboarding steps #185 — 🌋 Urgent
• Knowledge checks at onboarding milestones — gate phase progression, multiple-choice + short answer (Backend/Frontend, AI later) [Story]: Knowledge checks at onboarding milestones #128 — 🏕️ Medium
• Onboarding-blueprint freshness check — dedicated extra AI agent, criteria list (broken links, missing / renamed artefacts, content drift, blueprint inconsistencies), concrete remediation suggestions, no silent auto-overwrite (AI/Backend/Frontend) [Story]: Onboarding-blueprint freshness check (AI) #203 — 🏔️ High

B. Chat precision and summaries

• Source filter in the chat — docs / code / tickets + time range "latest" / "last 6 months" (Backend/Frontend/AI) [Story]: Source filter in the chat #104 — 🏔️ High
• Short summaries for long artifacts — "what changed" / key decisions, with citation, via the local LLM (Frontend/AI) [Story]: Short summaries for long artifacts #179 — 🏕️ Medium

C. Project management and second connector

• Project Management (Admin) — Admin project list + create / delete, user↔project mapping (default 1 user = 1 project), PM assignment, PM "Meine Projekte" view, User self-view; frontend on top of [Story]: Admin project management (create / delete projects + assign PMs) #197, with RBAC per [Story]: RBAC with Admin / User separation #93 (Frontend) [Feature]: Mobile View (onboarding & knowledge quality) #74 — priority TBD
• Jira CSV issue connector adapter — second source into the canonical artefact model, second priority behind the GitHub connector (Backend) [Story]: Jira CSV issue connector adapter #160 — 🏔️ High

D. Process & guidelines

• Vibe Coding — shared guidelines, tips & best practices — cross-team doc + team-specific docs in each repo (backend / frontend / AI), linked from README / CONTRIBUTING.md; English only (Docs, all teams) [Story]: Vibe Coding — shared guidelines, tips & best practices (cross-team + per-repo) #223 — 🌋 Urgent

Demo outcomes (Sprint Review 2026-07-15)

1. As a PM, drag a step in a team member's onboarding path to a new position; the new order is persisted, the progress bar stays correct, and the audit log shows who changed what and when.
2. As a new member, hit a milestone in the onboarding path; a short knowledge check appears, a wrong answer blocks the next phase with an explanation, and a correct answer unlocks it.
3. As a PM, run the freshness check on the role blueprints; the dedicated AI agent flags a step with a broken citation, a renamed artefact and a content-drift case between two role blueprints, and proposes a concrete remediation for each.
4. As a user in the chat, restrict the answer to "docs only / last 6 months" and ask a question; the answer is grounded only in the filtered sources.
5. As a user, open a long artefact and request a summary; the summary highlights "what changed" where version history is available, with a citation back to the source.
6. As an Admin, create a new project, assign a PM, map a user to the project; as the same user, log in and see the project + the assigned PM in the self-view; as the PM, see the project listed under "Meine Projekte".
7. (Stretch) Import a Jira CSV export; the tickets become queryable artefacts in the chat alongside the GitHub-ingested content, with provenance.
8. Open the cross-team Vibe Coding doc and the team-specific docs in each repo (backend / frontend / AI); all are linked from each repo's README and written in English.