claude-evolve 1.7.17 → 1.7.18

package/bin/claude-evolve-ideate

@@ -1019,20 +1019,18 @@ THE CSV FILE IS VERY LARGE (OVER 100,000 TOKENS). YOU WILL RUN OUT OF CONTEXT IF
 CRITICAL INSTRUCTIONS:
 1. Read ONLY the last 20-30 lines of the CSV to see the placeholder rows
 2. DO NOT ADD OR DELETE ANY ROWS - only EDIT the placeholder descriptions
-3. DO NOT CHANGE THE IDs - they are already correct
-4. Use the Edit tool to replace each PLACEHOLDER description with a real idea
-5. Each replacement should be one clear sentence describing a novel algorithmic approach"
+3. DO NOT CHANGE THE IDs - they are already correct ($required_ids_str)
+4. Use the Edit tool to replace EACH PLACEHOLDER text with a real algorithmic idea
+5. When editing, preserve the CSV structure: keep the ID and parent_id fields unchanged"
 
   if [[ -n "$existing_py_files" ]]; then
     prompt+="
-5. IMPORTANT: The following IDs already have Python files: $existing_py_files
+6. IMPORTANT: The following IDs already have Python files: $existing_py_files
    (This is informational only - use the IDs specified above)"
   fi
-  
+
   prompt+="
-6. Use the Edit or MultiEdit tool to APPEND exactly $count new rows AT THE END of the CSV file
-7. For each idea, create a row with: id,,description,,pending (empty parent_id for novel ideas)
-8. CRITICAL CSV FORMATTING RULES:
+7. CRITICAL CSV FORMATTING RULES:
    - ALWAYS wrap the description field in double quotes
    - If the description contains quotes, escape them by doubling them (\" becomes \"\")
    - Example: gen01-001,,\"Implement adaptive RSI thresholds based on volatility\",,pending
@@ -1171,13 +1169,14 @@ THE CSV FILE IS VERY LARGE (OVER 100,000 TOKENS). YOU WILL RUN OUT OF CONTEXT IF
 CRITICAL INSTRUCTIONS:
 1. Read ONLY the last 25 lines using the offset specified above
 2. DO NOT ADD OR DELETE ANY ROWS - only EDIT the placeholder descriptions
-3. DO NOT CHANGE THE IDs - they are already correct
-4. Use the Edit tool to replace each PLACEHOLDER description with a parameter tuning idea
-5. You may also change the parent_id field if needed to reference a different top performer
-6. Each description should focus on adjusting specific parameters - include current and new values
+3. DO NOT CHANGE THE IDs - they are already correct ($required_ids_str)
+4. Use the Edit tool to replace EACH PLACEHOLDER text with a parameter tuning idea
+5. When editing, preserve the CSV structure: keep the ID field unchanged
+6. You may change the parent_id field if needed to reference a different top performer
+7. Each description should focus on adjusting specific parameters - include current and new values
    Example: \"Lower rsi_entry from 21 to 18\" or \"Increase MA period from 20 to 50\"
-7. CRITICAL: When editing, preserve the CSV formatting with proper quoting
-8. DO NOT use any git commands (git add, git commit, git reset, etc.). Only modify the file directly."
+8. CRITICAL: When editing, preserve the CSV formatting with proper quoting
+9. DO NOT use any git commands (git add, git commit, git reset, etc.). Only modify the file directly."
 
   # Change to evolution directory so AI can access files
   local original_pwd=$(pwd)
@@ -1291,12 +1290,13 @@ THE CSV FILE IS VERY LARGE (OVER 100,000 TOKENS). YOU WILL RUN OUT OF CONTEXT IF
 CRITICAL INSTRUCTIONS:
 1. Read ONLY the last 25 lines using the offset specified above
 2. DO NOT ADD OR DELETE ANY ROWS - only EDIT the placeholder descriptions
-3. DO NOT CHANGE THE IDs - they are already correct
-4. Use the Edit tool to replace each PLACEHOLDER description with a structural modification idea
-5. You may also change the parent_id field if needed to reference a different top performer
-6. Each description should focus on architectural/structural changes
-7. CRITICAL: When editing, preserve the CSV formatting with proper quoting
-8. DO NOT use any git commands (git add, git commit, git reset, etc.). Only modify the file directly."
+3. DO NOT CHANGE THE IDs - they are already correct ($required_ids_str)
+4. Use the Edit tool to replace EACH PLACEHOLDER text with a structural modification idea
+5. When editing, preserve the CSV structure: keep the ID field unchanged
+6. You may change the parent_id field if needed to reference a different top performer
+7. Each description should focus on architectural/structural changes
+8. CRITICAL: When editing, preserve the CSV formatting with proper quoting
+9. DO NOT use any git commands (git add, git commit, git reset, etc.). Only modify the file directly."
 
   # Change to evolution directory so AI can access files
   local original_pwd=$(pwd)
@@ -1410,12 +1410,13 @@ THE CSV FILE IS VERY LARGE (OVER 100,000 TOKENS). YOU WILL RUN OUT OF CONTEXT IF
 CRITICAL INSTRUCTIONS:
 1. Read ONLY the last 25 lines using the offset specified above
 2. DO NOT ADD OR DELETE ANY ROWS - only EDIT the placeholder descriptions
-3. DO NOT CHANGE THE IDs - they are already correct
-4. Use the Edit tool to replace each PLACEHOLDER description with a crossover idea
-5. You may also change the parent_id field if needed (choose the primary parent)
-6. Each description should combine actual elements from 2+ top performers
-7. CRITICAL: When editing, preserve the CSV formatting with proper quoting
-8. DO NOT use any git commands (git add, git commit, git reset, etc.). Only modify the file directly."
+3. DO NOT CHANGE THE IDs - they are already correct ($required_ids_str)
+4. Use the Edit tool to replace EACH PLACEHOLDER text with a crossover idea
+5. When editing, preserve the CSV structure: keep the ID field unchanged
+6. You may change the parent_id field if needed (choose the primary parent)
+7. Each description should combine actual elements from 2+ top performers
+8. CRITICAL: When editing, preserve the CSV formatting with proper quoting
+9. DO NOT use any git commands (git add, git commit, git reset, etc.). Only modify the file directly."
 
   # Change to evolution directory so AI can access files
   local original_pwd=$(pwd)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "claude-evolve",
-  "version": "1.7.17",
+  "version": "1.7.18",
   "bin": {
     "claude-evolve": "./bin/claude-evolve",
     "claude-evolve-main": "./bin/claude-evolve-main",