Claude Memory
A built-in cross-session memory feature in Claude that allows personal information such as a user's occupation, preferences, and habits to be retained across multiple conversations. Users can manage memory entries manually, or Claude can generate and update them automatically.
Timeline (last 90 days)
A highly upvoted suggestion in the thread was to turn off the memory feature, because content unrelated to the prompt may have been loaded into the context and triggered the guardrail.
Claude's memory entries enter the prompt as additional information to help the model maintain coherence, but they can also cause context pollution: once historical snippets unrelated to the current request are recalled, they change the classifier's overall judgment of the input.
When encountering an inexplicable refusal, first turn off the memory feature and retry, which can rule out interference from context pollution.
All Facts (3)
A highly upvoted suggestion in the thread was to turn off the memory feature, because content unrelated to the prompt may have been loaded into the context and triggered the guardrail.
50%UnverifiedClaude's memory entries enter the prompt as additional information to help the model maintain coherence, but they can also cause context pollution: once historical snippets unrelated to the current request are recalled, they change the classifier's overall judgment of the input.
50%UnverifiedWhen encountering an inexplicable refusal, first turn off the memory feature and retry, which can rule out interference from context pollution.
50%