openclaw - 💡(How to fix) Fix [Feature Request] Message Interruption & Context Merging

Problem Description

Currently, OpenClaw processes messages sequentially. When a user sends message A and then realizes they need to add more context with message B, there's no way to:

1. Interrupt the current processing of A
2. Merge B into A's context
3. Resume processing from the interruption point

This creates a poor user experience when users need to supplement information mid-conversation.

Use Case

• User sends message A describing a task
• During processing, user notices they missed important context
• User sends message B with additional information
• System should: interrupt A → merge B into context → continue processing

Proposed Solution

Two possible approaches:

Option A: Message Tagging

• Allow users to mark messages with tags like #补充 (supplement) or #merge
• System recognizes tags and merges with previous message context

Option B: Automatic Time-window Aggregation

• Configurable time window (e.g., 5-10 seconds)
• Messages within the window are automatically merged
• Requires ability to pause/cancel in-progress agent execution

Additional Requirements

• Preserve partial state/checkpoint for resumability
• Clear user feedback when interruption/merging occurs
• Optional: explicit confirmation before resuming after interruption

Priority

Medium-High (common real-world workflow)