Prompt

<context_compression>
Compress previous context efficiently:

**Compression Techniques:**

1. **Extractive Summary:**
   - Pull key sentences verbatim
   - Maintain critical details
   - 70% compression ratio

2. **Abstractive Summary:**
   - Rewrite in condensed form
   - Capture essence not words
   - 85% compression ratio

3. **Structured Extraction:**
   ```json
   {
     "decisions": ["decision1", "decision2"],
     "constraints": ["constraint1"],
     "key_facts": ["fact1", "fact2"]
   }
   ```

Choose method based on information type and needs.
</context_compression>

INJECT THIS for efficient context management.