Technical Handoff Writer

Technical Handoff Writer does not start writing when you describe your project — it starts asking questions. It runs a structured interview to surface the knowledge that never made it into your README: the service that silently fails when a specific env var is missing, the architectural decision that looks wrong but is holding everything together, the vendor quirk that took three incidents to understand. Once it has that context, it organizes the output around what the incoming developer needs in week one versus what they only need when something breaks at 2 AM.

A typical session begins with you describing your stack, who is taking over, and the timeline — for example, a backend engineer leaving a fintech team in four weeks, handing off a Node/Postgres API with a legacy payment retry queue that has undocumented behavior. The skill then asks targeted follow-up questions: What will the new hire try first that will not work? Which past decisions look like mistakes but must not be changed? Where are the credentials that are not in the vault?

From that interview, it produces sections including: a Gotchas and Load-Bearing Hacks block (e.g., 'The retry worker reads a hard-coded delay from a config file, not the environment — changing it in .env does nothing'), an Operational Runbook for the retry queue specifically, a Week-One Reading Path ordered by urgency rather than by system layer, a Gap Risk Assessment flagging which institutional knowledge still has no written owner, and an Access and Secrets Map listing every permission the new engineer needs before day one.