Skip to content

Fix

Why Cursor 'Forgets' Context Mid-Task (and How to Drive It)

By The Field Academy Editorial TeamUpdated

Cursor 'forgets' because every model has a finite context window, and a long chat fills it with stale tokens that crowd out what matters. The fix is context hygiene: start a fresh chat for each task, give the agent only the files it needs, and restate constraints near the end of long sessions.

Why does Cursor forget what I told it?

Models read a fixed-size context window. As a chat grows — many files, long outputs, back-and-forth — older instructions get pushed toward the edges where the model weights them less, or fall out entirely. It looks like forgetting; it's really a full, noisy context.

This is a skill, not a defect

Managing context is the core craft of working with coding agents. Teams that treat it as a learnable workflow get dramatically more reliable results than teams that blame the model.

How do I fix it?

  1. 1Start a fresh chat per task. Don't reuse one endless thread across unrelated work.
  2. 2Scope context tightly. @-mention only the files in play; close irrelevant tabs.
  3. 3Restate the constraint when a session gets long ("remember: don't change the public API").
  4. 4Reset at ~70% context. When a thread feels heavy, summarize the state and start fresh.
  5. 5Use rules for durable conventions so you're not re-explaining them every chat.

Why does Cursor keep making the same mistake?

Two causes: the correction never made it into durable context (it lived in a now-buried message), or there's no verification step so the agent can't tell it failed. Put the rule in a .cursor/rules file, and give the agent a concrete way to check itself (a test or command) so it self-corrects instead of looping.

Frequently asked questions

What is Cursor's context limit?

It depends on the model you select — each has its own context window. Rather than chasing the largest window, keep context clean: the quality of what you include matters more than the quantity.

Should I start a new chat or keep one going?

Start a new chat per distinct task. Long-running threads accumulate stale context that degrades results; a fresh, tightly-scoped chat almost always performs better.

Do rules survive across chats?

Yes. .cursor/rules/*.mdc files are loaded automatically, so conventions persist across chats and teammates — the right home for anything you'd otherwise repeat.

Sources & last verified

Cursor ships frequently. Facts verified against primary sources on June 15, 2026.