fix: use async with for learning context manager

[zzstoatzz]

Summary

The learning SDK's context manager needs to be used with async with when wrapping async code. Using sync with creates a sync client, but the async interceptor expects an async client for memory operations.

This was causing the learning SDK to silently fail - no memory was being injected or captured because the sync client doesn't work properly in an async context.

Change

# Before (broken)
with learning(agent=f"slackbot-{user_context['user_id']}"):
    result = await create_agent(...).run(...)

# After (fixed)  
async with learning(agent=f"slackbot-{user_context['user_id']}"):
    result = await create_agent(...).run(...)

Test plan

• Deploy and verify memory persistence works across conversations

🤖 Generated with Claude Code

[Copilot]

Pull request overview

This PR fixes a critical bug in the slackbot example where the learning context manager from the agentic-learning SDK was being used incorrectly with synchronous with instead of asynchronous async with. Since the wrapped code uses await to run async operations, the async context manager is required for proper memory injection and capture.

Key changes:

• Changed with learning(...) to async with learning(...) in the slackbot's async agent run function
• Ensures memory persistence works correctly across conversations by using the proper async client

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.