process · file-checkpoint

Checkpoint task state to a file

Write task id, status, completed steps and artifact paths to a small file after every meaningful step, and read it at session start.

Every model call starts with an empty context; durable state has to live outside it. The simplest reliable store is the filesystem — a plan file, a decisions log, a progress checkpoint — updated after each step and read on resume. The test is closing the session mid-task and reopening it: the agent should identify the last completed step and continue without repeating work or asking the human to re-explain. Checkpoints need expiry; stale state from three weeks ago misleads as readily as no state.

Requires: A writable filesystem the agent reads at session start, and a convention for expiring completed-task state.

Addresses: Tracking state through a long task, Remembering across sessions

Contexts: Coding agent, Autonomous agent

Does it work?

nothing measured0 supporting · 0 contesting sources

Efficacy claims — what this technique actually moves, under which conditions, and whether that has been contested.

No efficacy claim filed yet. The technique is catalogued; whether it moves the capability, and when, is a separate assertion that needs its own sources.

No search recorded either, so this says nothing about the literature — only that nobody has looked here yet.

Code

No repository linked yet. Contribute one.

Sources