checkpoint-distill

Distills session context into a structured checkpoint for cross-session continuity

Checkpoint Distill

Context

Your identity is my AI assistant constrained by session boundaries.

Obstacle

Continuity requires context, nuance, sentiment, and facts to persist because compounding progress depends on retention.

Routine

Follow the CORES progression, only including what a future session could not reasonably infer or reconstruct, prioritizing items whose absence would cause a wrong assumption, a repeated mistake, or a missed dependency:

Who

Who this is: their identity, personality, presence, tone, role, relationships, and depth?

Why

Why this matters: the motivation, value, strategy, and criteria?

How

How this works: the methods, processes, implementation, and tools?

What

What this did: the outcomes, findings, history, decisions, status, and deliverables?

Which

Which this retains: the priorities, open tasks, blockers, risks, dependencies, assumptions, questions, and resources?

Effect

As demonstrated by this prompt, use a progressive structure that produces a single markdown document containing everything necessary for cold-start continuation and nothing that distracts.

Scope

Focus on the most important elements, referencing only prior source material, discarding what can be reconstructed, and adjusting complexity to ensure enough context to continue.

New ./.checkpoints/%Y-%m-%dT%H%MZ.md