AI Session Handoff — Shift Change for Working Sessions
End a working session like a shift change, not an abandonment: state captured, decisions logged, next step named — ready for the next session to pick up.
The flagship at full power: every decision, constraint, risk, assumption, and pending item of a complex project, transferred forensically as the new session's first message.
Some handoffs cannot afford triage: a complex rebuild with intertwined decisions, dual constraints, two open questions, named risks, and load-bearing assumptions — drop any one and the next session damages something. This is PROJECT STATE TRANSFER at full power: Forensic fidelity on a checkout-rebuild checkpoint, where everything detected travels, assumptions are first-class, the next step derives from the pending list, and the instructions enforce the complete contract — accept, preserve, question-don't-override, track, continue, and quote this package rather than reconstruct from memory.
Checkpoint everything
Forensic mode drops nothing — the full detected state becomes the canonical snapshot.
Make assumptions explicit
The idempotency window, the mobile-release dependency — given conditions travel as ASSUMPTIONS.
Enforce the full contract
Quote-don't-reconstruct plus question-don't-override — the strictest continuation rules in the tool.
Forensic fidelity drops nothing: every KEY DECISION, CONSTRAINT, OPEN QUESTION, RISK, and ASSUMPTION travels, and ASSUMPTIONS become first-class, like the 24-hour idempotency window. That completeness costs reading time, so the notFor points quick session boundaries to Compact. The context-handoff-builder assembles the package; you review and adjust the VERBATIM-extracted items before pasting it as the first message of the new chat.
The HANDOFF INSTRUCTIONS bind it: accept everything as established, preserve the key decisions without revisiting them, and if it believes a decision or assumption is wrong, raise it as an explicit question rather than override. It must also quote this package instead of reconstructing from memory. The context-handoff-builder writes these rules into the message you paste; enforcing them in the new session is on you.
The RECOMMENDED NEXT STEP is derived from the PENDING TASKS list, here "inventory reservation integration, the failure-path UX, and the load test at 2x peak", framed as a continuation rather than a fresh start. The HANDOFF INSTRUCTIONS tell the model to resume from it. The context-handoff-builder assembles this; you confirm the pending items are current before sending.
End a working session like a shift change, not an abandonment: state captured, decisions logged, next step named — ready for the next session to pick up.
When the window fills, don't retype the project from memory. A handoff package carries the decisions, constraints, and pending work into the new session as its first message.
Done work tells stories; open work needs carrying. The package separates completed from pending and resumes from the first open task.
Token budget planning for real workloads: how much of the window a transcript actually consumes, what is left for the answer, and how much headroom remains.
The "message too long" error has a structural fix: split at paragraph boundaries into sequenced chunks with wait rules, instead of pasting fragments and hoping.
What a converged conversation contains: a rejected voice, an accepted framing, a hard constraint, a format decision, and a final-version moment — each mapped to its prompt section.
Continue work in a new chat without losing context — decisions, constraints, and tasks extracted into a state package.
Carry a project into a new chat, model, or teammate without the context evaporating — capture the state, distill what's worth keeping, and rebuild it as durable context on the other side.