@nexus-cortex/server 4.26.0

package/.cortex/skills/pdf-documents/sections/11-form-filling.md

@@ -0,0 +1,241 @@
+# PDF Form Detection & Filling
+
+## Overview
+
+Detect AcroForm fields in existing PDF documents, inventory every field for the
+user, fill values programmatically with baked appearance streams, and deliver
+a completed PDF. Works with interactive AcroForm PDFs and provides a coordinate-overlay
+fallback for scanned/non-interactive forms.
+
+All processing via PyMuPDF. Install: `pip install pymupdf`
+
+---
+
+## Step 1: Install PyMuPDF (if needed)
+
+```bash
+pip install pymupdf
+```
+
+## Step 2: Detect and Inventory All Fields
+
+```python
+import fitz
+import json
+
+doc = fitz.open('form.pdf')
+
+# PyMuPDF widget field_type values:
+# 0=unknown, 1=PushButton, 2=CheckBox, 3=RadioButton,
+# 4=ListBox, 5=ComboBox, 6=Signature, 7=Text
+TYPE_NAMES = {0:'unknown', 1:'button', 2:'checkbox', 3:'radio',
+              4:'listbox', 5:'combobox', 6:'signature', 7:'text'}
+
+fields = []
+for page_num in range(len(doc)):
+    page = doc[page_num]
+    for widget in page.widgets():
+        field = {
+            'page': page_num + 1,
+            'index': len(fields),
+            'name': widget.field_name,
+            'type': TYPE_NAMES.get(widget.field_type, 'unknown'),
+            'type_id': widget.field_type,
+            'value': widget.field_value,
+            'rect': list(widget.rect),
+            'read_only': bool(widget.field_flags & 1),
+        }
+        if widget.field_type in (4, 5) and widget.choice_values:
+            field['options'] = widget.choice_values
+        if widget.field_type == 7 and widget.text_maxlen > 0:
+            field['max_length'] = widget.text_maxlen
+        fields.append(field)
+
+doc.close()
+
+print(f'Found {len(fields)} form fields across {len(doc)} pages')
+for f in fields:
+    ro = ' [READ-ONLY]' if f['read_only'] else ''
+    val = f' = "{f["value"]}"' if f['value'] else ''
+    opts = f' options: {f["options"]}' if 'options' in f else ''
+    print(f'  [{f["index"]:2d}] p{f["page"]} | {f["type"]:12} | {f["name"]}{val}{opts}{ro}')
+```
+
+Present the inventory to the user and collect values before filling.
+
+**IMPORTANT:** Fields with duplicate names (e.g., multiple "County" fields on the same
+form) must be addressed by index, not by name. The index is the widget's position
+in the `page.widgets()` list.
+
+## Step 3: Fill the Form
+
+```python
+import fitz
+
+fill_data = {
+    'Full Name': 'John A. Smith',
+    'Date of Birth': '03/15/1985',
+    'State': 'TX',
+    'I Agree': True,
+}
+
+# For duplicate field names, use index-based overrides:
+# fill_by_index = { 10: 'Jefferson', 12: 'Jefferson', 23: 'Jefferson' }
+
+doc = fitz.open('form.pdf')
+
+for page in doc:
+    widgets = list(page.widgets())
+    for i, widget in enumerate(widgets):
+        # Set font properties for baked appearance streams
+        if widget.field_type in (6, 7):  # Signature or Text
+            widget.text_fontsize = 0     # auto-fit to field rect
+            widget.text_color = (0, 0, 0)
+
+        # Index-based overrides first (for duplicate field names)
+        # if i in fill_by_index:
+        #     widget.field_value = fill_by_index[i]
+        #     widget.update()
+        #     continue
+
+        name = widget.field_name
+        if name not in fill_data:
+            continue
+
+        value = fill_data[name]
+
+        if widget.field_type == 2:  # checkbox
+            widget.field_value = 'Yes' if value else 'Off'
+        elif widget.field_type == 3:  # radio
+            widget.field_value = str(value)
+        elif widget.field_type == 6:  # signature — typed name
+            widget.field_value = str(value)
+        else:
+            widget.field_value = str(value)
+
+        widget.update()
+
+# ── CRITICAL: Bake appearance streams ──────────────────────────
+# Set NeedAppearances=false so ALL PDF viewers render the filled
+# values. Without this, many viewers (web, mobile, lightweight)
+# show blank fields — the #1 cause of "it filled but looks empty."
+cat_xref = doc.pdf_catalog()
+acro_ref = doc.xref_get_key(cat_xref, 'AcroForm')
+if acro_ref[0] == 'xref':
+    af_xref = int(acro_ref[1].split()[0])
+    doc.xref_set_key(af_xref, 'NeedAppearances', 'false')
+
+doc.save('form-filled.pdf', garbage=3, deflate=True)
+doc.close()
+print('Saved filled form with baked appearances')
+```
+
+### Why NeedAppearances Matters
+
+PDF forms have two rendering modes:
+- **NeedAppearances=true** — asks the viewer to regenerate visual appearances at
+  open time. Many viewers DON'T support this → fields appear blank.
+- **NeedAppearances=false** — PyMuPDF's `widget.update()` bakes an AP (appearance
+  stream) into each field. These render in ALL viewers.
+
+**Always set NeedAppearances=false after filling.** This is the difference between
+a form that works everywhere and one that looks empty in half of all viewers.
+
+## Step 4: Verify All Fields Filled
+
+```python
+import fitz
+
+doc = fitz.open('form-filled.pdf')
+filled = 0
+empty = []
+for page in doc:
+    for i, w in enumerate(page.widgets()):
+        v = w.field_value or ''
+        if v and v != 'Off':
+            filled += 1
+        elif w.field_type != 2:
+            empty.append(f'[{i}] {w.field_name}')
+        else:
+            filled += 1
+
+total = filled + len(empty)
+if empty:
+    print(f'WARNING: {len(empty)}/{total} fields empty: {empty}')
+else:
+    print(f'All {total} fields verified')
+doc.close()
+```
+
+## Step 5: Flatten (Optional — Lock Fields)
+
+```python
+import fitz
+
+doc = fitz.open('form-filled.pdf')
+
+for page in doc:
+    for widget in list(page.widgets()):
+        rect = widget.rect
+        value = widget.field_value or ''
+        if value and value != 'Off':
+            page.insert_text(
+                fitz.Point(rect.x0 + 2, rect.y1 - 2),
+                value, fontsize=10, fontname='helv',
+            )
+    page.clean_contents()
+
+doc.save('form-final.pdf', garbage=4, deflate=True)
+doc.close()
+```
+
+---
+
+## Fallback: Scanned / Non-Interactive Forms
+
+If the PDF has no AcroForm fields, overlay text at specific coordinates:
+
+```python
+import fitz
+
+doc = fitz.open('scanned-form.pdf')
+page = doc[0]
+
+# 1. Render as image to identify field locations
+pix = page.get_pixmap(dpi=150)
+pix.save('form-preview.png')
+
+# 2. Overlay text at identified positions (coordinates in PDF points, 72/inch)
+fields = [
+    (fitz.Point(150, 245), 'John A. Smith'),
+    (fitz.Point(150, 290), '03/15/1985'),
+    (fitz.Point(150, 335), '123 Main St, Dallas, TX 75201'),
+]
+
+for point, text in fields:
+    page.insert_text(point, text, fontsize=11, fontname='helv', color=(0, 0, 0))
+
+doc.save('scanned-filled.pdf')
+doc.close()
+```
+
+---
+
+## Field Type Reference
+
+| Type ID | Type | Fill Pattern | Notes |
+|---------|------|-------------|-------|
+| 2 | Checkbox | `'Yes'` / `'Off'` | Some PDFs use custom on_state values |
+| 3 | Radio | Option label text | Set to the option's export value |
+| 4 | Listbox | Option value | Must match one of `choice_values` |
+| 5 | Combobox | Option value or free text | Depends on field flags |
+| 6 | Signature | `/s/ Name` | Accepts text for typed signatures |
+| 7 | Text | Any string | Set `text_fontsize=0` for auto-fit |
+
+## Tips
+
+- Always inventory fields first — present to user before filling
+- Watch for duplicate field names — use index-based filling when names collide
+- `text_fontsize = 0` tells PyMuPDF to auto-scale text to fit the field rect
+- **Always set NeedAppearances=false** — this is the #1 cause of blank fields
+- Signature fields (type 6) accept typed text like `/s/ Full Name` for e-signatures

package/.cortex/skills/pptx/SKILL.md

@@ -0,0 +1,90 @@
+---
+name: pptx
+description: >
+  Create, read, and edit PowerPoint presentations (.pptx). Use whenever the
+  deliverable is a slide deck — building presentations from content or data,
+  editing existing decks, extracting slide text, or deck-to-PDF conversion. Do
+  not trigger when the requested output is a document or web page.
+metadata:
+  short-description: "Create, read, and edit PowerPoint (.pptx) decks"
+  author: "nexus-cortex"
+---
+
+# PPTX — Presentations
+
+Build and edit .pptx decks with open-source tooling. Decks are a LAYOUT medium:
+the most common failure isn't broken files, it's unreadable slides — too much
+text, overflowing frames, invisible contrast. Design for the back of the room.
+
+## Tooling (pick by job)
+
+| Job | Tool | Notes |
+|-----|------|-------|
+| Create / edit programmatically | `python-pptx` | Slides, layouts, placeholders, text frames, images, tables, charts, notes |
+| Markdown → deck quickly | `pandoc` | `pandoc slides.md -o deck.pptx` (`---`-separated slides; `--reference-doc` for branding) |
+| Extract slide text | `python-pptx` walk | Iterate slides → shapes → `text_frame` |
+| pptx → PDF | LibreOffice headless | `soffice --headless --convert-to pdf deck.pptx --outdir out/` |
+
+Install on demand: `pip install python-pptx`.
+
+## Core patterns (python-pptx)
+
+```python
+from pptx import Presentation
+from pptx.util import Inches, Pt, Emu
+from pptx.dml.color import RGBColor
+
+prs = Presentation()                          # or Presentation("existing.pptx")
+prs.slide_width  = Inches(13.333)             # 16:9 — set BEFORE adding slides
+prs.slide_height = Inches(7.5)
+
+title_slide = prs.slides.add_slide(prs.slide_layouts[0])   # 0=title, 1=title+content
+title_slide.shapes.title.text = "Q3 Results"
+title_slide.placeholders[1].text = "Engineering review — 2026"
+
+slide = prs.slides.add_slide(prs.slide_layouts[1])
+slide.shapes.title.text = "Highlights"
+body = slide.placeholders[1].text_frame
+body.text = "Latency down 40%"                # first bullet
+p = body.add_paragraph(); p.text = "Cost per request halved"; p.level = 1
+
+slide.shapes.add_picture("chart.png", Inches(7), Inches(1.5), width=Inches(5.5))
+slide.notes_slide.notes_text_frame.text = "Mention the cache work here."
+prs.save("deck.pptx")
+```
+
+Editing an existing deck: iterate `prs.slides` → `slide.shapes`; only shapes
+with `shape.has_text_frame` carry text. Edit at the run level
+(`text_frame.paragraphs[i].runs[j].text`) to preserve formatting. To match the
+deck's branding, ALWAYS build new slides from `prs.slide_layouts` of the same
+file rather than a fresh `Presentation()`.
+
+## Slide design rules (what makes decks fail review)
+
+- **One idea per slide; ≤ 6 bullets; ≤ 10 words per bullet.** Prose belongs in
+  the speaker notes, not the slide.
+- **Minimum 18pt body text** (titles 28pt+). If content doesn't fit at 18pt,
+  split the slide — never shrink the font to fit.
+- **Check contrast** — light text needs a dark fill behind it; never place text
+  straight onto a busy image without a scrim shape.
+- **Charts as images** (matplotlib → PNG → `add_picture`) are more reliable than
+  native chart XML for complex visuals; use native `pptx.chart` only for simple
+  bar/line/pie that must stay editable.
+- Position with explicit `Inches()` — overlapping autolayout shapes are the #1
+  source of "looks broken" decks.
+
+## Verification (mandatory)
+
+1. **Re-open** with `Presentation("deck.pptx")` — corruption throws.
+2. **Assert structure**: slide count, expected titles, image/shape presence.
+3. **Render-check when layout matters**: convert to PDF via LibreOffice and
+   inspect the pages (text overflow and overlap are invisible in the XML — only
+   a render shows them).
+
+## Limits to know
+
+- SmartArt, animations, and transitions are not supported by `python-pptx` —
+  preserved if present, but not creatable/editable. Say so; don't fake it.
+- Embedded video/audio: insert as a linked file or warn — round-trip is fragile.
+- Fonts must exist on the rendering machine; stick to widely-available families
+  (Calibri, Arial, Helvetica) unless the user supplies brand fonts.

package/.cortex/skills/resume-analyst/SKILL.md

@@ -0,0 +1,373 @@
+---
+name: resume-analyst
+description: Analyze, improve, and create professional resumes and CVs. Use when the user needs help with resume review, ATS optimization, content improvement, or creating a new resume from scratch.
+allowed-tools: Read, Write, Grep, Glob
+---
+
+# Resume & CV Analyst
+
+A comprehensive skill for analyzing, improving, and creating professional resumes and CVs with ATS (Applicant Tracking System) optimization.
+
+## When to Use This Skill
+
+- User wants to review or improve an existing resume
+- User needs help creating a new resume
+- User asks for ATS optimization advice
+- User wants feedback on resume structure or content
+- User needs to tailor a resume for a specific job posting
+
+## Instructions
+
+### Phase 1: Analysis (Existing Resume)
+
+If the user has an existing resume:
+
+1. **Locate the Resume File**
+   - Use Glob to find resume files: `*.pdf`, `*.docx`, `*.txt`, `resume.*`, `cv.*`
+   - Common locations: current directory, `~/Documents/`, `~/Desktop/`
+   - Ask user for file location if not found
+
+2. **Read and Parse Content**
+   - Use Read tool to access the file
+   - For PDF: Use Bash with `pdftotext` or `strings` if needed
+   - For Word: Use Bash with `antiword` or `docx2txt` if available
+   - Extract all text content for analysis
+
+3. **Analyze Resume Structure**
+   Check for these essential sections:
+   - ✅ Contact Information (name, email, phone, location, LinkedIn/portfolio)
+   - ✅ Professional Summary/Objective
+   - ✅ Work Experience (with dates, company, role, achievements)
+   - ✅ Education (degree, institution, graduation date)
+   - ✅ Skills (technical and soft skills)
+   - ✅ Optional: Certifications, Projects, Publications, Awards
+
+4. **ATS Optimization Check**
+   Evaluate for ATS compatibility:
+   - ❌ **Avoid**: Graphics, tables, columns, headers/footers, text boxes, images
+   - ✅ **Use**: Simple formatting, standard fonts (Arial, Calibri, Times New Roman)
+   - ✅ **Include**: Keywords from job description
+   - ✅ **Format**: Standard section headings, consistent date formats
+   - ✅ **File Type**: .docx or .txt preferred over PDF for ATS
+
+5. **Content Quality Assessment**
+   Evaluate each work experience entry:
+   - **Achievement-Focused**: Quantifiable results (increased X by Y%, managed $Z budget)
+   - **Action Verbs**: Strong verbs (Led, Implemented, Optimized, not "Responsible for")
+   - **STAR Format**: Situation, Task, Action, Result
+   - **Relevance**: Aligned with target role
+   - **Clarity**: Clear, concise bullet points (1-2 lines each)
+
+### Phase 2: Feedback & Recommendations
+
+Provide structured feedback:
+
+1. **Overall Assessment** (1-10 score with rationale)
+   - ATS Compatibility: __/10
+   - Content Quality: __/10
+   - Structure & Format: __/10
+   - Professional Impact: __/10
+
+2. **Critical Issues** (Must Fix)
+   - Missing essential sections
+   - ATS-incompatible formatting
+   - Lack of quantifiable achievements
+   - Typos or grammatical errors
+   - Outdated or irrelevant content
+
+3. **Improvement Opportunities** (Should Fix)
+   - Weak action verbs
+   - Vague descriptions
+   - Missing keywords
+   - Inconsistent formatting
+   - Poor space utilization
+
+4. **Enhancement Suggestions** (Nice to Have)
+   - Additional relevant skills
+   - Project showcases
+   - Leadership examples
+   - Industry certifications
+
+### Phase 3: Content Improvement
+
+For each work experience entry, suggest improved versions:
+
+**Before**:
+```
+Responsible for managing team projects and coordinating with clients.
+```
+
+**After**:
+```
+• Led cross-functional team of 8 engineers to deliver 15+ client projects on time, achieving 95% customer satisfaction
+• Coordinated with 20+ enterprise clients, resulting in $2M+ annual contract renewals
+```
+
+**Guidelines for Improvements**:
+- Start with strong action verbs
+- Include specific numbers and metrics
+- Show impact and results
+- Use industry-standard terminology
+- Keep bullet points concise (1-2 lines)
+
+### Phase 4: Resume Generation (New Resume)
+
+If creating a new resume from scratch:
+
+1. **Gather Information**
+   Ask the user for:
+   - Personal details (name, contact, location)
+   - Target role and industry
+   - Work history (companies, dates, responsibilities, achievements)
+   - Education background
+   - Skills and certifications
+   - Notable projects or accomplishments
+
+2. **Choose Format**
+   - **Chronological**: Best for steady career progression
+   - **Functional**: Best for career changers or gaps
+   - **Combination**: Best for experienced professionals
+
+3. **Generate Resume Content**
+   Use this structure:
+
+```
+[FULL NAME]
+[Email] | [Phone] | [City, State] | [LinkedIn URL] | [Portfolio/GitHub]
+
+PROFESSIONAL SUMMARY
+[2-3 sentences highlighting years of experience, key expertise, and value proposition]
+
+PROFESSIONAL EXPERIENCE
+
+[Company Name] | [Location]
+[Job Title] | [Month Year] - [Month Year]
+• [Achievement-focused bullet point with metrics]
+• [Achievement-focused bullet point with metrics]
+• [Achievement-focused bullet point with metrics]
+
+[Repeat for each role]
+
+EDUCATION
+
+[Degree] in [Field] | [University Name] | [Graduation Year]
+• [Honors, GPA if >3.5, relevant coursework]
+
+SKILLS
+
+Technical: [List of relevant technical skills]
+Tools: [List of relevant tools and platforms]
+Soft Skills: [List of relevant soft skills]
+
+CERTIFICATIONS (if applicable)
+• [Certification Name], [Issuing Organization], [Year]
+```
+
+4. **Write to File**
+   - Save as `resume_[name]_[date].txt` or `.md`
+   - Use Write tool to create the file
+   - Inform user they can convert to PDF/Word as needed
+
+### Phase 5: Job-Specific Tailoring
+
+If user has a specific job posting:
+
+1. **Extract Job Requirements**
+   - Read job description
+   - Identify required skills, experience, qualifications
+   - Note key terminology and phrases
+   - Identify company values and culture
+
+2. **Match Resume Content**
+   - Highlight relevant experience
+   - Incorporate job-specific keywords
+   - Reorder bullet points for relevance
+   - Add/emphasize matching skills
+   - Adjust professional summary
+
+3. **Customize Cover Letter** (if requested)
+   - Reference specific job requirements
+   - Connect experience to role needs
+   - Show cultural fit
+   - Include call-to-action
+
+## ATS Optimization Checklist
+
+Use this checklist for ATS compatibility:
+
+- [ ] Simple, clean formatting (no tables, columns, graphics)
+- [ ] Standard section headings (EXPERIENCE, EDUCATION, SKILLS)
+- [ ] Standard fonts (Arial, Calibri, Times New Roman, 10-12pt)
+- [ ] .docx or .txt file format
+- [ ] No headers/footers (put contact info in body)
+- [ ] No text boxes or images
+- [ ] Keywords from job description included
+- [ ] Consistent date formatting (MM/YYYY or Month YYYY)
+- [ ] Acronyms spelled out first time (e.g., "Search Engine Optimization (SEO)")
+- [ ] Standard bullet points (•, -, or ◦)
+- [ ] File name format: FirstName_LastName_Resume.docx
+
+## Common Issues & Solutions
+
+### Issue: "Responsible for..."
+**Problem**: Passive language that doesn't show impact
+**Solution**: Start with action verbs and add results
+```
+❌ Responsible for customer support
+✅ Resolved 200+ customer inquiries monthly with 98% satisfaction rating
+```
+
+### Issue: No Metrics
+**Problem**: Vague, unmeasurable claims
+**Solution**: Add specific numbers
+```
+❌ Improved team efficiency
+✅ Improved team efficiency by 35% through process automation, reducing turnaround time from 5 days to 3 days
+```
+
+### Issue: Too Long or Too Short
+**Problem**: Resume length not appropriate for experience level
+**Solution**:
+- Entry-level (0-5 years): 1 page
+- Mid-level (5-10 years): 1-2 pages
+- Senior-level (10+ years): 2 pages maximum
+
+### Issue: Outdated Skills
+**Problem**: Lists irrelevant or outdated technologies
+**Solution**: Focus on current, in-demand skills for target role
+
+### Issue: Employment Gaps
+**Problem**: Unexplained gaps in work history
+**Solution**:
+- Use years instead of months if gap is small
+- Include relevant activities (freelance, courses, volunteering)
+- Address positively in cover letter if needed
+
+## Industry-Specific Guidance
+
+### Tech/Engineering
+- Emphasize: Technical skills, programming languages, frameworks, projects
+- Include: GitHub/portfolio links, open-source contributions
+- Format: Clean, modern, data-driven
+- Keywords: Agile, CI/CD, APIs, cloud platforms, version control
+
+### Business/Finance
+- Emphasize: Results, revenue impact, cost savings, strategic initiatives
+- Include: Certifications (CPA, CFA, MBA), leadership experience
+- Format: Professional, conservative
+- Keywords: P&L, ROI, stakeholder management, financial modeling
+
+### Creative/Design
+- Emphasize: Portfolio, design process, tools, creative achievements
+- Include: Portfolio link prominently, design awards
+- Format: Can be more creative but still ATS-friendly
+- Keywords: UX/UI, Adobe Creative Suite, design thinking, user research
+
+### Healthcare
+- Emphasize: Certifications, patient care, compliance, specialized skills
+- Include: Licenses, credentials, patient outcomes
+- Format: Professional, detail-oriented
+- Keywords: EMR systems, HIPAA, patient-centered care, clinical protocols
+
+### Marketing/Sales
+- Emphasize: Campaign results, growth metrics, brand awareness, revenue
+- Include: Successful campaigns, client portfolios, certifications
+- Format: Results-driven, dynamic
+- Keywords: ROI, lead generation, conversion rates, digital marketing, CRM
+
+## Examples
+
+### Example 1: Software Engineer
+
+**Before**:
+```
+Software Developer at TechCorp (2020-2023)
+- Wrote code for various projects
+- Fixed bugs
+- Worked with team members
+```
+
+**After**:
+```
+Software Engineer | TechCorp | San Francisco, CA | Jan 2020 - Dec 2023
+• Architected and deployed 12+ microservices handling 10M+ daily requests with 99.9% uptime
+• Reduced application load time by 40% through code optimization and caching strategies
+• Mentored 5 junior developers, conducting code reviews and technical workshops
+• Technologies: Python, Django, PostgreSQL, Redis, AWS, Docker, Kubernetes
+```
+
+### Example 2: Marketing Manager
+
+**Before**:
+```
+Marketing Manager at StartupXYZ (2019-2023)
+- Managed marketing campaigns
+- Worked on social media
+- Helped increase brand awareness
+```
+
+**After**:
+```
+Marketing Manager | StartupXYZ | Austin, TX | Mar 2019 - Present
+• Drove 250% increase in qualified leads through multi-channel campaigns, generating $5M in pipeline
+• Managed $500K annual marketing budget, optimizing spend to achieve 35% reduction in CAC
+• Built marketing team from 2 to 8 members, establishing processes and campaign frameworks
+• Increased social media engagement by 180% across LinkedIn, Twitter, and Facebook platforms
+```
+
+### Example 3: Project Manager
+
+**Before**:
+```
+Project Manager at ConsultCo (2018-2023)
+- Managed client projects
+- Coordinated with teams
+- Delivered projects on time
+```
+
+**After**:
+```
+Senior Project Manager | ConsultCo | Chicago, IL | Jun 2018 - Aug 2023
+• Led 25+ enterprise consulting engagements ($100K-$2M each) for Fortune 500 clients with 100% on-time delivery
+• Managed cross-functional teams of 15+ consultants across 3 time zones, delivering complex digital transformation projects
+• Improved client satisfaction scores from 7.2 to 9.1/10 through proactive communication and risk management
+• Achieved 98% project profitability rate by implementing agile methodologies and resource optimization
+```
+
+## Key Reminders
+
+1. **Quantify Everything**: Numbers, percentages, dollar amounts, timeframes
+2. **Be Specific**: Replace vague terms with concrete examples
+3. **Show Impact**: Focus on results, not just responsibilities
+4. **Use Keywords**: Match language from job descriptions
+5. **Keep it Current**: Remove outdated skills and old experiences (>10-15 years)
+6. **Proofread Thoroughly**: No typos, consistent formatting, proper grammar
+7. **Optimize for ATS**: Simple format, standard headings, keyword-rich
+8. **Tell a Story**: Show career progression and growth
+9. **Be Honest**: Never fabricate experience or qualifications
+10. **Keep it Concise**: Every word should add value
+
+## Resources to Reference
+
+- **Action Verbs List**: Achieved, Accelerated, Accomplished, Administered, Analyzed, Architected, Built, Championed, Collaborated, Coordinated, Created, Delivered, Demonstrated, Designed, Developed, Drove, Enhanced, Established, Exceeded, Executed, Expanded, Generated, Implemented, Improved, Increased, Initiated, Launched, Led, Managed, Maximized, Optimized, Orchestrated, Organized, Pioneered, Reduced, Resolved, Spearheaded, Streamlined, Strengthened, Transformed
+
+- **Metrics to Include**:
+  - Percentages (increased by X%)
+  - Dollar amounts ($X revenue, $Y budget)
+  - Time saved (reduced from X to Y)
+  - Team size (managed X people)
+  - Customer/user counts (served X customers)
+  - Project counts (delivered X projects)
+  - Quality metrics (98% satisfaction, 99.9% uptime)
+
+---
+
+## Final Output Format
+
+Always provide:
+1. **Analysis Summary** (if reviewing existing resume)
+2. **Scored Assessment** (ATS, Content, Structure, Impact)
+3. **Prioritized Recommendations** (Critical, Important, Enhancement)
+4. **Specific Improvements** (before/after examples)
+5. **Updated Resume Draft** (if requested)
+6. **Next Steps** (actions user should take)