How Much of a Bloated MEMORY.md Actually Loads
An engineer's auto memory MEMORY.md has grown to 340 lines (roughly
30KB) after months of a long-running project. They want to know
exactly what happens at the start of the next brand-new session.
Which statement correctly describes how much of this file Claude Code loads into context at session start?
The correct answer is "Only the first 200 lines or 25KB, whichever comes first."
MEMORY.md is loaded up to a hard cutoff: the first 200 lines, or the
first 25KB, whichever limit is hit first — content past that point
simply isn't read into context at session start. At 340 lines, this
file is already well past the 200-line limit, so a real chunk of it —
potentially including entries the engineer actually needs — is
silently invisible on every new session until the file is shortened.
The distractors misdescribe the mechanism: there is a real, hard cutoff, so the whole file does not load regardless of size; Claude Code doesn't stop reading the file altogether once it's over the limit (it still loads up to the cutoff, and separately, writes to an over-limit file still succeed but come back with an error nudging Claude to rewrite the index) — it isn't a full read-lockout; and the cutoff is positional (the first N lines/KB of the file as written), not based on which entries were edited most recently.
Share this question