← ClaudeAtlas

knowledge-transferlisted

Use when a colleague is leaving a project and must hand over knowledge, or when someone new joins a project that contains a handover/ folder. Two phases: export (analyze project + Claude memories, produce a committed handover package: non-technical onboarding doc + filtered portable memories) and import (verify those memories against current code, install them for the new colleague). Trigger on /knowledge-transfer, 'handover', 'knowledge transfer', 'passaggio di consegne', 'onboard a new colleague', or when the user mentions taking over / leaving a project.
ferdinandobons/knowledge-transfer · ★ 0 · AI & Automation · score 75
Install: claude install-skill ferdinandobons/knowledge-transfer
# /knowledge-transfer Transfer project knowledge between colleagues: a short non-technical onboarding document plus the AI's accumulated project memories, packaged inside the project repo — so the newcomer starts oriented and their AI starts informed. ## Usage ``` /knowledge-transfer # auto-detect mode (see below) /knowledge-transfer export # outgoing colleague: build the handover package /knowledge-transfer import # new colleague: verify + install memories, read the doc ``` ## Mode detection (no argument) 1. If `handover/manifest.json` exists in the project root → announce import mode. 2. Otherwise → announce export mode. State the detected mode in one line and proceed. The user can override at any time. ## Contracts Read these before producing anything: - `references/package-format.md` — package structure and manifest schema - `references/privacy-filter.md` — what may and may not be exported - `references/onboarding-template.md` — structure, tone, and length of the document --- ## Export phase Run from the project root by the colleague leaving the project. ### 1. Analyze the project - Read README, CLAUDE.md (if present), the top-level structure, entry points, and main config files. Use subagents for large codebases rather than reading everything inline. - Identify: what the project does, the 3–7 main components, and 2–3 end-to-end logical flows worth narrating. - Detect the project's dominant language (README, docs, recent commit messa