aiblueprint-cli 1.4.22 → 1.4.24

package/claude-code-config/skills/meta-skill-creator/references/prompting-integration.md

@@ -0,0 +1,363 @@
+<overview>
+How to integrate expert prompting techniques from the create-prompt skill into skill creation. Every skill should apply proven prompting principles for maximum effectiveness.
+</overview>
+
+<core_principle>
+Skills are specialized prompts that get loaded on-demand. All prompting best practices from create-prompt apply to skill creation, with emphasis on:
+
+- Clear, specific instructions
+- XML tags for structure
+- Examples when format matters
+- Success criteria defined
+- Context management for long-running tasks
+</core_principle>
+
+<prompting_techniques_for_skills>
+<technique name="be_clear_and_direct">
+**Apply to skills by**:
+
+- State exactly what Claude should do
+- Avoid ambiguous language ("try to", "maybe", "generally")
+- Use "Always..." or "Never..." instead of "Should probably..."
+- Provide specific output format requirements
+
+**Example in SKILL.md**:
+
+❌ **Vague**:
+```xml
+<workflow>
+Try to extract the data from the PDF. If it works, save it somewhere.
+</workflow>
+```
+
+✅ **Clear**:
+```xml
+<workflow>
+1. **Extract data**: Use `pdfplumber.open()` to read PDF
+2. **Parse content**: Call `extract_text()` on each page
+3. **Save output**: Write to JSON file with UTF-8 encoding
+</workflow>
+```
+</technique>
+
+<technique name="use_xml_tags">
+**Apply to skills by**:
+
+- Use pure XML structure (no markdown headings)
+- Separate sections with semantic tags
+- Wrap data in descriptive tags
+- Define clear boundaries
+
+Skills already enforce this through required structure. See [xml-tag-guide.md](xml-tag-guide.md).
+</technique>
+
+<technique name="few_shot_examples">
+**Apply to skills by**:
+
+Provide 2-4 input/output pairs when output format matters:
+
+**Example in SKILL.md**:
+```xml
+<examples>
+<example number="1">
+<task>Extract product names from invoice PDF</task>
+<code>
+```python
+with pdfplumber.open("invoice.pdf") as pdf:
+    text = pdf.pages[0].extract_text()
+    products = re.findall(r'Product: (.+)', text)
+```
+</code>
+<output>
+['Widget A', 'Gadget B', 'Tool C']
+</output>
+</example>
+
+<example number="2">
+<task>Extract table from financial report</task>
+<code>
+```python
+with pdfplumber.open("report.pdf") as pdf:
+    table = pdf.pages[1].extract_table()
+```
+</code>
+<output>
+[['Quarter', 'Revenue'],
+ ['Q1', '$1.2M'],
+ ['Q2', '$1.5M']]
+</output>
+</example>
+</examples>
+```
+</technique>
+
+<technique name="chain_of_thought">
+**Apply to skills by**:
+
+For complex tasks, add explicit reasoning steps:
+
+**Example in SKILL.md**:
+```xml
+<workflow>
+<step_1>
+**Analyze PDF structure**: First, inspect the PDF to understand layout and content type. Use `pdfplumber.open()` to examine pages.
+</step_1>
+
+<step_2>
+**Determine extraction method**: Based on structure, choose between:
+- `extract_text()` for plain text
+- `extract_table()` for tabular data
+- Custom parsing for complex layouts
+</step_2>
+
+<step_3>
+**Extract content**: Apply chosen method to each page, handling errors gracefully.
+</step_3>
+
+<step_4>
+**Validate output**: Check for encoding issues, missing data, or formatting problems before saving.
+</step_4>
+</workflow>
+```
+</technique>
+
+<technique name="success_criteria">
+**Apply to skills by**:
+
+Always include `<success_criteria>` tag with specific, measurable criteria:
+
+**Example in SKILL.md**:
+```xml
+<success_criteria>
+The extraction is successful when:
+
+- All pages processed without errors
+- Text encoding is correct (no mojibake)
+- Output file created with expected structure
+- Data completeness verified (no missing sections)
+- File size is reasonable (not empty or corrupted)
+</success_criteria>
+```
+</technique>
+
+<technique name="context_management">
+**Apply to skills by**:
+
+For long-running tasks, add state tracking guidance:
+
+**Example in SKILL.md**:
+```xml
+<workflow>
+<state_tracking>
+For large batches, track progress:
+
+```python
+import json
+from pathlib import Path
+
+progress_file = Path("progress.json")
+
+# Load previous progress
+if progress_file.exists():
+    progress = json.loads(progress_file.read_text())
+else:
+    progress = {"processed": [], "failed": []}
+
+# Process files
+for pdf_file in pdf_files:
+    if str(pdf_file) in progress["processed"]:
+        continue  # Skip already processed
+
+    try:
+        extract_pdf(pdf_file)
+        progress["processed"].append(str(pdf_file))
+    except Exception as e:
+        progress["failed"].append({"file": str(pdf_file), "error": str(e)})
+
+    # Save progress after each file
+    progress_file.write_text(json.dumps(progress, indent=2))
+```
+</state_tracking>
+</workflow>
+```
+</technique>
+</prompting_techniques_for_skills>
+
+<skill_specific_patterns>
+<pattern name="explicit_workflow">
+**When to use**: Multi-step processes
+
+Instead of vague instructions, provide explicit numbered steps:
+
+❌ **Vague**:
+```xml
+<objective>
+Process PDFs to extract data and save results.
+</objective>
+```
+
+✅ **Explicit**:
+```xml
+<objective>
+Extract structured data from PDF files through a 4-step process: analyze, extract, validate, save.
+</objective>
+
+<workflow>
+1. **Analyze PDF structure**: Inspect layout and content type
+2. **Extract content**: Apply appropriate extraction method
+3. **Validate output**: Check completeness and encoding
+4. **Save results**: Write to structured format (JSON/CSV)
+</workflow>
+```
+</pattern>
+
+<pattern name="anti_patterns_section">
+**When to use**: Common mistakes exist
+
+Help Claude avoid mistakes by showing bad vs good examples:
+
+```xml
+<anti_patterns>
+<pitfall name="incorrect_encoding">
+❌ **Bad**: Assuming UTF-8 encoding
+
+```python
+text = pdf.pages[0].extract_text()
+file.write(text)  # May fail with encoding errors
+```
+
+**Why bad**: PDFs may use various encodings
+
+✅ **Good**: Specify encoding explicitly
+
+```python
+text = pdf.pages[0].extract_text()
+file.write(text, encoding='utf-8', errors='replace')
+```
+</pitfall>
+</anti_patterns>
+```
+</pattern>
+
+<pattern name="validation_guidance">
+**When to use**: Output verification is important
+
+Provide specific validation steps:
+
+```xml
+<validation>
+<output_verification>
+Verify extracted data quality:
+
+1. **Completeness**: Check all expected sections present
+2. **Accuracy**: Sample-check values against PDF
+3. **Encoding**: Verify special characters preserved
+4. **Structure**: Confirm JSON/CSV structure valid
+</output_verification>
+
+<automated_checks>
+```python
+def validate_output(output_file):
+    """Validate extracted data."""
+    with open(output_file) as f:
+        data = json.load(f)
+
+    assert "text" in data, "Missing text field"
+    assert len(data["text"]) > 0, "Empty text"
+    assert data["page_count"] > 0, "Invalid page count"
+
+    return True
+```
+</automated_checks>
+</validation>
+```
+</pattern>
+
+<pattern name="progressive_examples">
+**When to use**: Skills have multiple complexity levels
+
+Start simple, build complexity:
+
+```xml
+<examples>
+<example number="1" difficulty="beginner">
+<task>Extract text from single page</task>
+<code>
+```python
+import pdfplumber
+with pdfplumber.open("simple.pdf") as pdf:
+    text = pdf.pages[0].extract_text()
+    print(text)
+```
+</code>
+</example>
+
+<example number="2" difficulty="intermediate">
+<task>Extract all pages and save to file</task>
+<code>
+```python
+with pdfplumber.open("document.pdf") as pdf:
+    all_text = []
+    for page in pdf.pages:
+        all_text.append(page.extract_text())
+
+    with open("output.txt", "w") as f:
+        f.write("\n\n".join(all_text))
+```
+</code>
+</example>
+
+<example number="3" difficulty="advanced">
+<task>Extract tables with custom settings and error handling</task>
+<code>
+```python
+with pdfplumber.open("report.pdf") as pdf:
+    tables = []
+    for i, page in enumerate(pdf.pages):
+        try:
+            table = page.extract_table({
+                "vertical_strategy": "lines",
+                "horizontal_strategy": "text"
+            })
+            if table:
+                tables.append({"page": i+1, "data": table})
+        except Exception as e:
+            print(f"Page {i+1} failed: {e}")
+
+    with open("tables.json", "w") as f:
+        json.dump(tables, f, indent=2)
+```
+</code>
+</example>
+</examples>
+```
+</pattern>
+</skill_specific_patterns>
+
+<integration_checklist>
+When creating a skill, ensure you've applied these prompting techniques:
+
+- [ ] **Clarity**: Instructions are specific and unambiguous
+- [ ] **Structure**: Pure XML structure used throughout
+- [ ] **Examples**: Included when output format matters
+- [ ] **Success criteria**: Specific, measurable criteria defined
+- [ ] **Workflow**: Multi-step processes broken down explicitly
+- [ ] **Anti-patterns**: Common mistakes shown and avoided
+- [ ] **Validation**: Output verification guidance provided
+- [ ] **Context management**: State tracking for long-running tasks
+- [ ] **Progressive complexity**: Examples range from simple to advanced
+- [ ] **Error handling**: Edge cases and failures addressed
+</integration_checklist>
+
+<reference_create_prompt>
+For detailed prompting techniques, consult the create-prompt skill:
+
+- Clarity principles
+- XML structure patterns
+- Few-shot examples
+- Chain of thought reasoning
+- Context management strategies
+- Anti-patterns to avoid
+
+Use create-prompt when writing SKILL.md content to ensure expert-level prompting quality.
+</reference_create_prompt>