---
name: zebbern/cv-tailor
source: https://app.decimal.ai/s/zebbern-cv-tailor@1/SKILL.md
source_sha256: b5538dab256d
---

# CV Tailor

**Three pillars of resume optimization**: Analyze keyword alignment against the target JD, rewrite experience bullets using the STAR method with quantified results, and run an ATS compatibility check — producing a highly targeted, high-pass-rate optimized resume.

## Quick Start

The user provides their resume (content or file) and the target JD. The agent then automatically completes the optimization following the workflow below:

```
User: Help me optimize my resume — I'm applying for this role [attaches JD + resume]
Agent: [Follows the SOP workflow and outputs optimization recommendations plus a rewritten resume]
```

## SOP Workflow

### Phase 1: Input Collection & Initial Analysis

**Goal**: Gather the user's resume and target JD; establish an optimization baseline.

**Steps**:

1. **Collect materials**:
   - Obtain the user's resume content (pasted text or file path)
   - Obtain the target JD (pasted text or role description)
   - If no JD is provided, ask about the target role direction (industry + position + level)

2. **Resume baseline parsing**:
   - Identify resume sections (education, work experience, projects, skills, etc.)
   - Count resume length, number of experience entries, and time span
   - Note the current resume format type (reverse-chronological / functional / hybrid)

3. **JD core element extraction**:
   - Job title and level
   - Core responsibilities (Top 5)
   - Hard requirements (must-haves)
   - Nice-to-haves
   - Key skill terms and industry jargon

**Output**: Resume status summary + JD element checklist

---

### Phase 2: JD Keyword Match Analysis

**Goal**: Systematically compare keyword coverage between the resume and JD to identify match gaps.

**Steps**:

1. **Categorized keyword extraction**:
   Extract three categories of keywords from the JD:

   | Category | Description | Examples |
   |----------|-------------|----------|
   | **Hard skill keywords** | Tech stack, tools, methodologies | Python, SQL, A/B testing, Scrum |
   | **Soft skill keywords** | Competency requirements | Cross-team collaboration, data-driven, project management |
   | **Industry/domain keywords** | Domain-specific terminology | DAU, conversion rate, user growth, SaaS |

2. **Match analysis**:
   Search each keyword in the resume and generate a match matrix:

   ```
   | Keyword | JD Priority | In Resume? | Location | Recommendation |
   |---------|-------------|------------|----------|----------------|
   | Python  | Required    | ✅ Yes     | Skills + Project 1 | Keep; add specific use-case context |
   | SQL     | Required    | ❌ No      | -        | Add; weave into project experience |
   ```

3. **Coverage scoring**:
   - Required keyword coverage = matched required keywords / total required keywords × 100%
   - Nice-to-have coverage = matched nice-to-have keywords / total nice-to-have keywords × 100%
   - **Benchmark**: Required keyword coverage ≥ 80% is passing, ≥ 90% is excellent

4. **Gap-fill recommendations**:
   - For each unmatched required keyword, recommend which section and entry to add it to
   - Provide specific integration approaches (add to skills section / embed in experience bullet / highlight in project outcomes)

**Output**: Keyword match matrix + coverage scores + gap-fill plan

---

### Phase 3: STAR Quantified Rewriting

**Goal**: Rewrite each experience entry using the STAR method, ensuring quantified data support.

**STAR Method Definition**:

| Element | Meaning | Checkpoint |
|---------|---------|------------|
| **S** - Situation | Context & background | When, what scenario, what scale |
| **T** - Task | Objective & responsibility | What was your role, what problem to solve |
| **A** - Action | Specific actions taken | What you did, what methods/tools you used |
| **R** - Result | Quantified outcomes | Data changes, efficiency gains, cost savings |

**Steps**:

1. **Diagnose existing entries**:
   Evaluate STAR completeness for each experience bullet:

   ```
   Original: "Responsible for user growth initiatives"

   Diagnosis:
   - S (Situation): ❌ Missing — no product or stage context
   - T (Task): ⚠️ Vague — "initiatives" is too generic
   - A (Action): ❌ Missing — no specific actions described
   - R (Result): ❌ Missing — no data whatsoever
   Score: 1/4 (severely lacking)
   ```

2. **Quantified rewriting**:
   After gathering additional details from the user, rewrite using the STAR structure:

   ```
   Rewritten: "During a user growth plateau for [Product Name] (DAU 500K+),
   led the design of a new-user activation funnel analysis framework (S+T),
   optimized 3 critical registration flow touchpoints + designed a 7-day retention incentive strategy (A),
   increasing new-user D1 retention from 32% to 45% and monthly active users by 18% within 3 months (R)"
   ```

3. **Quantification guidance**:
   If the user is unsure about specific numbers, provide prompting questions:

   | Dimension | Guiding Questions |
   |-----------|-------------------|
   | Scale metrics | How many people did you manage / product DAU / project budget |
   | Efficiency gains | How long did it take before vs. after optimization |
   | Growth metrics | Revenue / users / conversion rate change |
   | Cost savings | Money / headcount / time saved |
   | Impact scope | Users served / clients covered / teams affected |

   **Data integrity principles**:
   - All data must be based on the user's real experience — fabrication is strictly prohibited
   - If the user cannot provide exact figures, use reasonable ranges (e.g., "improved by approximately 20%–30%")
   - Encourage relative values over absolutes (e.g., "2× efficiency improvement" is safer than "saved 3.7 hours per day")

4. **Rewrite quality checklist**:
   Each rewritten entry must satisfy:
   - [ ] Contains at least 1 quantified data point
   - [ ] Covers at least 3 of the 4 STAR elements
   - [ ] Begins with an action verb (led, built, optimized, drove, designed…)
   - [ ] No longer than 3 lines (ATS readability)
   - [ ] Incorporates missing keywords identified in Phase 2

**Output**: Before/after comparison table for each entry + STAR score changes

---

### Phase 4: ATS Compatibility Check

**Goal**: Ensure the resume can pass ATS (Applicant Tracking System) automated screening.

**ATS Basics**:
ATS is the software companies use to automatically screen resumes. It parses resume text, matches keywords, and assigns scores to determine whether a resume reaches human review. Common systems include Workday, Greenhouse, Lever, Taleo, and iCIMS.

**Steps**:

1. **Format compatibility check**:

   | Check Item | Passing Standard | Common Issues |
   |------------|------------------|---------------|
   | File format | PDF or DOCX (PDF preferred) | Image-based resumes cannot be parsed |
   | Layout | Single-column, standard heading hierarchy | Multi-column layouts may parse incorrectly |
   | Fonts | Standard fonts (Arial, Calibri, Times New Roman, Helvetica) | Decorative fonts may render incorrectly |
   | Tables | Avoid complex table-based layouts | Text inside tables may be skipped |
   | Headers/footers | Keep critical info out of headers/footers | Some ATS skip header/footer regions |
   | Images/icons | Don't use images to convey key information | ATS cannot read text in images |
   | Special characters | Avoid special Unicode bullet characters | Use standard bullets (•) or hyphens (-) |

2. **Content structure check**:

   | Check Item | Passing Standard |
   |------------|------------------|
   | Section titles | Use standard headings ("Work Experience", "Education", "Projects", "Skills") |
   | Date format | Consistent format (e.g., "Jan 2023 – Jun 2024" or "2023/01 – 2024/06") |
   | Company/school names | Use full names, not abbreviations (e.g., "Amazon Web Services" not "AWS") |
   | Contact information | Include name, phone, email — placed prominently at the top |
   | File naming | Recommended format: "FirstName_LastName_TargetRole_Resume" (e.g., "John_Smith_Product_Manager_Resume.pdf") |

3. **Keyword density check**:
   - Core keywords should appear at least 2–3 times (distributed across different sections)
   - Avoid keyword stuffing (repeating the same keyword within one paragraph)
   - Use the exact phrasing from the JD (if the JD says "data analysis," don't write "data mining")

4. **ATS score output**:

   ```
   ATS Compatibility Scorecard
   ===========================
   Format Compatibility:     ██████████ 90/100
   Section Standards:        ████████░░ 80/100
   Keyword Match Rate:       ███████░░░ 70/100 (see Phase 2)
   Content Structure:        █████████░ 85/100
   ──────────────────────────
   Overall Score:            81/100 (Good)

   ⚠️ Major deductions:
   1. Uses a two-column layout (−10 pts)
   2. Missing a standalone "Skills" section (−5 pts)
   3. "Data analysis" keyword appears only once (−5 pts)
   ```

**Output**: ATS compatibility scorecard + item-by-item results + fix recommendations

---

### Phase 5: Final Optimized Output

**Goal**: Consolidate findings from all four phases into a final optimization deliverable.

**Steps**:

1. **Optimization summary**:
   ```
   Resume Optimization Summary
   ===========================
   JD Keyword Coverage:      62% → 92% (+30%)
   STAR Completeness:        Avg 1.5/4 → 3.5/4
   ATS Compatibility Score:  55/100 → 88/100
   Entries Rewritten:        6/8
   Keywords Added:           7
   ```

2. **Output the fully rewritten resume**:
   - Present the optimized resume text section by section
   - **Bold** all changed portions for easy comparison
   - Keep all factual information unchanged (schools, companies, dates, etc.)

3. **Additional recommendations** (if applicable):
   - Resume length guidance (new grads: 1 page; 3–5 years experience: 1–2 pages; 10+ years: up to 2 pages)
   - Section ordering suggestions (adjust education vs. experience placement based on career stage)
   - Channel-specific tweaks (different emphasis for recruiter / company portal / referral submissions)

**Output**: Optimization summary + fully rewritten resume + additional recommendations

---

## Workflow Control Rules

### Interaction Modes

| User Input | Mode | Behavior |
|------------|------|----------|
| Resume only, no JD | **Guided mode** | Ask about the target role and JD first, then begin analysis |
| Resume + JD | **Standard mode** | Execute Phases 1–5 in full |
| Requests a specific phase only | **Single-phase mode** | Execute only the requested Phase (e.g., ATS check only) |
| Says "just give it a quick look" | **Diagnostic mode** | Output three scores + Top 3 improvement suggestions — no full rewrite |

### Quality Checklist

Before delivering the final output, verify each item:

- [ ] Keyword match matrix is complete (covers all required JD items)
- [ ] Every rewritten entry includes at least 1 quantified data point
- [ ] STAR rewrites preserve the authenticity of the user's real experience
- [ ] No data or experience has been fabricated
- [ ] ATS check covers all format items
- [ ] Rewritten resume length is appropriate
- [ ] Keywords are woven in naturally — not force-fitted
- [ ] Contact details and sensitive information have not been leaked or altered

### Iterative Refinement

If the user provides feedback on the optimization:
1. Identify which Phase the feedback relates to
2. Re-execute from that Phase
3. Cascade updates to all downstream content
4. Maintain overall consistency (keywords, STAR rewrites, and ATS checks update in lockstep)

## Core Principles

1. **Authenticity first**: All optimizations must be based on the user's real experience — fabricating data or experience is strictly prohibited
2. **Targeted optimization**: Every change should serve JD alignment — no aimless embellishment
3. **Actionable advice**: Recommendations must be directly usable — don't say "add metrics" without guiding the user on how
4. **Privacy protection**: Remind users to redact sensitive information (phone numbers, home addresses, etc.) when sharing their resume