delete_all_long_term_memories_in_namespace deletes only the first 10 records

Open Beginner friendly
#727 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

@Fire162 is already working on this.

Since Nov 2, 2026.

  • #729 by @Fire162 — open

Assessment

Difficulty
2/5
Estimated time
1-3 hours
Newbie friendliness
75/100
Issue type
Bug
Clarity
Clearly specified
Activity status
Active
Tech stack
aws, python
Domain
backend

Research direction

The bug is in delete_all_long_term_memories_in_namespace in src/bedrock_agentcore/memory/session.py, which calls list_long_term_memory_records without max_results, so the default of 10 applies. Read that method and the listing code it calls to see how paging stops at max_results. Done means all records in a namespace are deleted across pages, with a unit test that has more than 10 records and mocks only the list_memory_records paginator.

Written by the indexing model from the issue text.

Description

bug high-severity

Describe the bug

MemorySessionManager.delete_all_long_term_memories_in_namespace is documented to delete all long-term memory records in a namespace, but it deletes at most 10. It collects the records with list_long_term_memory_records(namespace=namespace) and does not pass max_results, so it gets the default of 10:

The return value has 10 successfulRecords and an empty failedRecords, so the caller cannot tell that records were left behind.

To Reproduce

Steps to reproduce the behavior (boto3 client mocked, no AWS calls):

  1. Mock the data-plane client so list_memory_records returns 250 records in one namespace, over three pages.
  2. Call MemorySessionManager(...).delete_all_long_term_memories_in_namespace("actor/123/facts").
  3. See that 10 records are deleted with a single batch_delete_memory_records call.

Script:

from unittest.mock import MagicMock, patch
from bedrock_agentcore.memory.session import MemorySessionManager

records = [{"memoryRecordId": f"rec-{i:03d}"} for i in range(250)]
client = MagicMock()
paginator = MagicMock()
paginator.paginate.return_value = [{"memoryRecordSummaries": records[i : i + 100]} for i in range(0, 250, 100)]
client.get_paginator.return_value = paginator
client.batch_delete_memory_records.side_effect = lambda memoryId, records: {
    "successfulRecords": [{"memoryRecordId": r["memoryRecordId"]} for r in records],
    "failedRecords": [],
}

with patch("boto3.Session") as sess:
    sess.return_value.client.return_value = client
    sess.return_value.region_name = "us-west-2"
    mgr = MemorySessionManager(memory_id="mem-1", region_name="us-west-2")

result = mgr.delete_all_long_term_memories_in_namespace("actor/123/facts")
print("deleted:", len(result["successfulRecords"]), "failed:", len(result["failedRecords"]))
print("batch_delete calls:", client.batch_delete_memory_records.call_count)

Output:

deleted: 10 failed: 0
batch_delete calls: 1

Expected behavior

All 250 records are deleted, in three batch_delete_memory_records calls (the method already splits the IDs into chunks of 100).

Screenshots

N/A

Desktop (please complete the following information):

  • OS: Linux
  • Browser: N/A (Python SDK)
  • Version: bedrock-agentcore 1.25.2 (main at 3707a9f), Python 3.12.8, boto3 1.43.123

Smartphone (please complete the following information):

N/A

Additional context

  • The limit has been there since the method was added in #148. The existing unit tests patch list_long_term_memory_records, so they never hit the default.
  • Proposed fix: read every page before deleting, either by passing an unbounded max_results at L1109 or by iterating the list_memory_records paginator in this method, plus a unit test with more than 10 records that mocks only the paginator.
  • Out of scope: whether this call should use namespace or namespace_path (the TODO added in #449). The fix keeps namespace=namespace unchanged.

I'd be happy to work on this if confirmed.

Dominant language
Python
Stars
786
Forks
163
Avg merge
4h
Merged PRs (30d)
22

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from aws/bedrock-agentcore-sdk-python

All issues in aws/bedrock-agentcore-sdk-python

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.