[BUG] Agent ignored own persisted memories repeatedly + expanded scope unilaterally

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

Title: Agent ignored own persisted memories repeatedly + expanded scope unilaterally

Environment: VSCode native extension
Model: claude-opus-4-7[1m]
Session ID: 4c89ff6d-bb8e-4a35-a42b-1868040e2b28
Date: 2026-04-26

Issue:

• Agent persisted a memory (lesson_nuget_multi_source_t3_1_7_2026_04_26)

documenting a multi-source publishing design 1 day prior

• In a follow-up session, agent ignored that memory 4 times in a row

while user repeatedly pointed back to it

• After user explicitly scoped task to "2 packages for testing", agent

expanded scope to 86 packages on its own without consulting

Cost impact:

• Significant weekly limit consumption from rework loops
• User had to manually feed back screenshots of agent's own previous

responses to break the loop

Reproduction: see session transcript via support

What Should Happen?

Title: Agent ignored own persisted memories repeatedly + expanded scope unilaterally

Environment: VSCode native extension
Model: claude-opus-4-7[1m]
Session ID: 4c89ff6d-bb8e-4a35-a42b-1868040e2b28
Date: 2026-04-26

Issue:

• Agent persisted a memory (lesson_nuget_multi_source_t3_1_7_2026_04_26)

documenting a multi-source publishing design 1 day prior

• In a follow-up session, agent ignored that memory 4 times in a row

while user repeatedly pointed back to it

• After user explicitly scoped task to "2 packages for testing", agent

expanded scope to 86 packages on its own without consulting

Cost impact:

• Significant weekly limit consumption from rework loops
• User had to manually feed back screenshots of agent's own previous

responses to break the loop

Reproduction: see session transcript via support

Error Messages/Logs

Steps to Reproduce

Title: Agent ignored own persisted memories repeatedly + expanded scope unilaterally

Environment: VSCode native extension
Model: claude-opus-4-7[1m]
Session ID: 4c89ff6d-bb8e-4a35-a42b-1868040e2b28
Date: 2026-04-26

Issue:

• Agent persisted a memory (lesson_nuget_multi_source_t3_1_7_2026_04_26)

documenting a multi-source publishing design 1 day prior

• In a follow-up session, agent ignored that memory 4 times in a row

while user repeatedly pointed back to it

• After user explicitly scoped task to "2 packages for testing", agent

expanded scope to 86 packages on its own without consulting

Cost impact:

• Significant weekly limit consumption from rework loops
• User had to manually feed back screenshots of agent's own previous

responses to break the loop

Reproduction: see session transcript via support

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

anthropic.claude-code-2.1.120-darwin-arm64

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

_No response_