Menu
Analyzes job descriptions and generates tailored resumes that highlight relevant experience, skills, and achievements to maximize interview chances
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Run AI coding agents on disposable repo clones. The agent works on a clone โ it can't touch your real repo. An optional container (Podman or Docker) provides build/test isolation. You review the diff and decide what (if anything) to apply.
Create Deezer playlists programmatically from any query โ similar artists, genre mixes, festival lineups, mood-based collections. Four-tier data pipeline: Deezer public REST API + Last.fm scrobble data for discovery, GQL Pipe API for smart mixes and playlist creation, web search for subjective curation. Uses ARL cookie auth โ no OAuth app required.
Generate compact AI-readable context maps from codebases โ tools like codesight, repomix, agentic-context that pre-compute project structure to save tokens in AI coding sessions.
Research-focused query handling with multi-source synthesis, citations, and Obsidian persistence. Like a self-hosted Perplexity/Vane but CLI-native. Best for quick-to-medium lookups using Kagi. Use when the user asks factual questions, needs citations, or wants a direct answer โ not a full research report (use deep-research) or social sentiment (use last30days). Triggers on: research, look into, what's the latest on, compare, explain, investigate.
Write articles, guides, blog posts, tutorials, newsletter issues, research reports, and deep research outputs in a distinctive voice derived from supplied examples or brand guidance. Use when the user wants polished written content longer than a paragraph, deep research on a topic, or a research report โ especially when voice consistency, structure, and credibility matter. Triggers on: 'write an article', 'research report', 'deep research', 'long-form', 'blog post', 'guide', 'newsletter', 'white paper', 'research paper'.
Systematic research methodology for major consumer durables (appliances, HVAC, power tools, outdoor equipment) and smart garden/outdoor devices (bird feeder cameras, bird baths, smart outdoor gadgets). Emphasis on real reliability data, failure mode analysis, and head-to-head comparison. Use when the user asks to research, review, compare, or evaluate major purchases where longevity and repair risk matter, or when researching smart bird feeders, bird bath cameras, and similar connected outdoor devices. Triggers on: washer/dryer, refrigerator, dishwasher, HVAC, furnace, AC, generator, power tool, appliance reviews, appliance reliability, which [appliance] to buy, compare models, bird feeder camera, bird bath camera, smart garden devices.
| name | tailored-resume-generator |
| description | Analyzes job descriptions and generates tailored resumes that highlight relevant experience, skills, and achievements to maximize interview chances |
See references/role-reframing.md for reusable reframing patterns (Support โ SE, Support โ CS/TAM, Support โ Infrastructure/Cloud, Support โ Technical Writer, Support โ CX Leadership) and application question drafting strategies.
See references/application-answer-template.md for the "Why this company + fit" answer template and tone guidance.
Use the reusable build script: scripts/md_to_docx.py Handles all standard resume markdown โ formatted DOCX. No need to rewrite the conversion each time. See the script header for prerequisites.
Provide a job description and your background information:
I'm applying for this job:
[paste job description]
Here's my background:
- 5 years as software engineer at TechCorp
- Led team of 3 developers on mobile app project
- Expert in Python, JavaScript, React
- Computer Science degree from State University
Upload or paste your current resume along with the job description:
Please tailor my resume for this position:
Job Description:
[paste job description]
My Current Resume:
[paste resume content]
When changing industries or roles:
I'm transitioning from marketing to product management.
Here's the job I'm applying for:
[paste job description]
My transferable experience:
- 7 years in digital marketing
- Led cross-functional teams
- Managed product launches
- Data analysis and A/B testing
User Request:
I need a tailored resume for this job:
Senior Data Analyst Position
Requirements:
- 5+ years experience in data analysis
- Expert in SQL, Python, and data visualization tools
- Experience with A/B testing and statistical analysis
- Strong communication skills to present findings
- Healthcare industry experience preferred
My background:
- Data Analyst at RetailCo (2019-2024)
- Created dashboards using Tableau and Power BI
- Built Python scripts for data automation
- Worked with marketing team on campaign analysis
- Business Analytics degree
- Also did 1 year internship at HealthPlus clinic
Generated Output:
# JOHN DOE
Email: john.doe@email.com | Phone: (555) 123-4567 | Location: City, State
## PROFESSIONAL SUMMARY
Results-driven Data Analyst with 5+ years of experience leveraging SQL, Python, and advanced
visualization tools to deliver actionable insights. Proven track record in statistical analysis,
A/B testing, and cross-functional collaboration. Healthcare industry experience with strong
ability to communicate complex data findings to stakeholders.
## TECHNICAL SKILLS
- **Data Analysis & Programming**: SQL, Python, Statistical Analysis, A/B Testing
- **Visualization Tools**: Tableau, Power BI, Advanced Dashboard Development
- **Data Management**: Data Automation, ETL Processes, Database Optimization
- **Business Intelligence**: Campaign Analytics, Performance Metrics, KPI Tracking
## PROFESSIONAL EXPERIENCE
**Data Analyst** | RetailCo | 2019 - 2024
- Designed and implemented 50+ SQL queries and Python automation scripts, reducing manual
data processing time by 60%
- Conducted rigorous A/B testing and statistical analysis for marketing campaigns,
improving campaign ROI by 35%
- Built interactive Tableau and Power BI dashboards for executive team, enabling
data-driven decision making
- Presented analytical findings to cross-functional stakeholders including marketing,
operations, and executive leadership
- Collaborated with marketing team to analyze customer behavior patterns and optimize
targeting strategies
**Data Analysis Intern** | HealthPlus Clinic | 2018 - 2019
- Supported healthcare analytics initiatives by analyzing patient data and operational metrics
- Gained exposure to healthcare industry data management and HIPAA compliance requirements
- Assisted in creating reports for clinical staff to improve patient care workflows
## EDUCATION
**Bachelor of Science in Business Analytics**
State University | Graduated 2018
## KEY ACHIEVEMENTS
- Reduced data processing time by 60% through Python automation
- Improved marketing campaign ROI by 35% through A/B testing and optimization
- Created data visualization system now used across entire organization
Additional Recommendations:
When a user requests resume tailoring:
When the user provides a URL, always extract and check the actual job title before proceeding. Users often paste a URL from a prior conversation or search session โ the job at that link may differ from what they discussed. If the extracted title doesn't match the role context, flag it immediately and confirm before tailoring.
Example: User says "tailor my resume for this Technical Support Manager job" but the URL points to a Solutions Engineer listing. Tailoring for the wrong role wastes a full cycle.
Job Description Analysis:
Candidate Background:
Extract and prioritize:
Create a mental map of:
For each job requirement:
Professional Summary (3-4 lines):
Technical/Core Skills Section:
Professional Experience:
Apply these frameworks to transform weak bullets into achievement-focused statements:
X-Y-Z Formula (Google Method):
"Accomplished [X] as measured by [Y] by doing [Z]"
Example:
โ BEFORE: "Managed social media accounts"
โ
AFTER: "Grew Instagram following by 250% (5K to 17.5K) by implementing daily content calendar and influencer partnerships"
STAR Method (condensed for bullets):
[Situation] + [Task] + [Action] + [Result]
Example:
"Inherited underperforming sales team (S) with 65% quota attainment. Tasked with improving performance within Q1 (T). Implemented new training program and revised commission structure (A). Achieved 92% quota attainment by Q2, generating $1.8M additional revenue (R)."
Power Verbs by Category:
Rule: Every bullet should have at least ONE number.
Metric Categories:
When exact numbers aren't available:
Quantification Templates:
Before/After: "Improved [X] from [before] to [after], resulting in [Y]% improvement"
Scale: "[Verb] [number] [things], resulting in [impact]"
Volume+Impact: "Processed [number] [items] per [time], achieving [quality metric]"
When user is changing careers/industries, use these translation patterns:
Universal Transferable Skills:
Industry Translation Examples:
Key Principle: Translate experience into the target industry's language while maintaining truthfulness.
For technical roles (SWE, PM, Data, DevOps):
Technical Bullet Formula:
[Action Verb] + [Technical What] + [Scale/Impact] + [Technology Used]
Example:
โ
"Architected microservices migration from monolith, reducing deployment time from 2 hours to 15 minutes and enabling independent team deployments"
โ
"Optimized PostgreSQL queries and implemented Redis caching, reducing API latency by 60% (from 500ms to 200ms) for 100K daily active users"
Technical Metrics to Include:
Technical Skills Section Organization:
What to Include for Tech Roles:
Education:
Optional Sections (if applicable):
ATS Compatibility Checklist (from Resume ATS Optimizer):
File Format:
Font & Formatting:
Section Headers (use standard, recognizable headers):
Contact Information (NOT in header/footer):
John Smith
email@example.com | (555) 123-4567 | LinkedIn: linkedin.com/in/johnsmith
San Francisco, CA
Keyword Optimization:
Match Score Calculation:
Match Score = (Keywords Matched / Total Required Keywords) ร 100
Target: 80%+ for strong match
ATS Red Flags to Avoid:
Format Options:
Resume Structure Guidelines:
After presenting the tailored resume, offer:
Strengths Analysis:
Gap Analysis:
Interview Preparation Tips:
Cover Letter Hooks:
This is a distinct workflow from resume generation. Use it when the user has an upcoming interview and wants to map their existing experience to the job description requirement-by-requirement, with scripted talking points and gap mitigation.
~/Documents/Resumes/, ~/Desktop/, or ~/Downloads/ for .docx, .pdf, or .md files. Ask the user to confirm if found.pandoc first. If unavailable, use python-docx via terminal (not execute_code, due to venv import path issues). Read all paragraphs.pdftotext or PyPDF2 in terminal.read_file.For each distinct requirement or responsibility in the JD:
Output format for each mapping:
**[Number]. "[JD requirement text]"**
Your talking point:
"[1-2 sentence interview-ready narrative]"
Evidence from your resume:
- "[exact resume bullet]"
- "[exact resume bullet]"
Gaps to address:
[bridging language or none]
**Salary context**: Glassdoor estimates $105Kโ$170K. OpenRouter advertises "top-of-market" packages.
**Your anchor**: $160Kโ$170K base (you exceed the 3โ5 year requirement significantly).
---
**1. "Own the technical resolution... first line of defense for code-level debugging"**
Your talking point:
"I've spent the last four years as the primary escalation resource between Support and Engineering. I reproduce, isolate, and either resolve or hand off fully scoped bugs to Engineering with root cause identified."
Evidence from your resume:
- "Act as primary escalation resource for complex customer-reported issues, coordinating resolution between Support and Engineering"
- "Write and troubleshoot PowerShell and Bash scripts for endpoint management automation, directly impacting product debugging and issue reproduction"
Gaps to address:
None โ this is a core strength.
---
**2. "Experience with Datadog, Cloudflare logs, or GCP Cloud Logging"**
Your talking point:
"I haven't used Datadog specifically, but I live in Grafana and Mode daily for observability. I'm confident I can transfer those skills quickly."
Evidence from your resume:
- "Leverage Mode and Grafana for data analysis and visualization to quantify issue scope, track trends, and inform cross-functional decisions"
Gaps to address:
No direct Datadog/Cloudflare experience. Frame as adjacent-tool proficiency + fast learner.
Ask if user wants to:
When a job application includes custom questions (common on Greenhouse, Lever, Ashby), draft answers alongside the resume. This is a natural extension of the tailoring work โ you already have the JD analysis and the candidate's background mapped. See references/application-answer-template.md for the "Why this company + fit" template.
Workflow:
Common question types and strategies:
references/application-answer-template.md for templates.~/Documents/Resumes/portfolio/. Host on GitHub repo to simultaneously demonstrate Docs-as-Code skills. See references/application-answer-template.md for full strategy.LinkedIn extraction fails 100% of the time โ web_extract, Firecrawl, and stealth mode all return empty. See references/linkedin-extraction-fallback.md for the proven fallback chain (search jobright.ai, Indeed, Lensa mirrors). Jobright.ai is the gold standard: structured data with responsibilities, qualifications, salary, and company context.
Always extract and verify the JD before tailoring. Job postings get filled, redirected, or replaced. When the user provides a URL:
Some aggregator sites (Virtual Vocations, Lensa, etc.) hide the company name and full JD behind a subscription. Don't give up โ the data is often in the page source as JSON-LD structured data:
document.querySelector('script[type="application/ld+json"]')?.textContenthiringOrganization object in the JSON โ specifically hiringOrganization.name. This field is required by schema.org/JobPosting and is rarely stripped even when the visual UI is gated."hiringOrganization": {"@type": "Organization", "name": "Acme Corp."}If the page uses Next.js, check document.querySelector('script#__NEXT_DATA__')?.textContent for the same data in a different format.
Fallback: Search for the job URL's unique ID or text snippet on other job boards (ZipRecruiter, Indeed, LinkedIn) โ aggregators often mirror listings that are more fully visible elsewhere.
Common mismatches:
Do:
Don't:
scripts/md_to_docx.py โ do NOT write a custom build script. The inline sample script below is incomplete and has known bugs. The canonical script handles all edge cases correctly. Copy it to /tmp/md_to_docx.py and run it: source /tmp/.venv/bin/activate && python3 /tmp/md_to_docx.py input.md output.docx. Only write a custom script if the canonical one fails for a specific reason.#/## header prefixes: The canonical script now strips # prefixes automatically, but the SAFEST approach is to write section headers as plain ALL CAPS text without any # prefix (e.g., PROFESSIONAL SUMMARY not ## PROFESSIONAL SUMMARY). The isupper() detection fails on lines starting with # because the # and space aren't uppercase. If you must use # headers, verify the script version includes the stripping logic.**Title** | Company | Location | Date have 4 pipe-separated parts. The canonical script renders all parts on one line with _add_rich() per part. If writing a custom script, do NOT assume exactly 3 parts โ join parts[2:] back together for the subtitle line, or the date gets silently dropped.- **Category:** items) should render as plain paragraphs with a bold category name, NOT as List Bullet items. The script detects these by leading - + ** + : within first 60 chars. If your markdown uses a different format (e.g., **Category:** items without the leading dash), they'll fall through to a different handler.| line. If you're writing a custom script, handle this by consuming consecutive text lines after a pipe-delimited entry.The Hermes sandbox terminal uses a venv that lacks pip and python-docx. Use this two-stage workflow:
# Use write_file to save as .md first โ this lets you iterate and review
content = """
NAME | Contact | Location
# (full resume content)
"""
write_file(path="~/Documents/Resumes/tailored_company.md", content=content)
Preferred: Use the canonical script. Copy it to a temp location and run:
cd /tmp && uv venv .venv && source .venv/bin/activate && uv pip install python-docx lxml
cp ~/.hermes/skills/tailored-resume-generator/scripts/md_to_docx.py /tmp/
python3 /tmp/md_to_docx.py input.md ~/Documents/Resumes/tailored_company.docx
Only if the canonical script fails: Create a temp venv and write a custom conversion script. The Hermes system Python is managed by uv and rejects pip install --system.
# โ ๏ธ LEGACY INLINE SAMPLE โ use scripts/md_to_docx.py instead (see Stage 2 above)
# This sample is illustrative only and has known bugs with 4-part job lines and ** markers.
# save as build_resume.py, then (with /tmp/.venv activated): python3 build_resume.py
from docx import Document
from docx.shared import Pt, Inches, RGBColor
from docx.enum.text import WD_ALIGN_PARAGRAPH
doc = Document()
style = doc.styles['Normal']
style.font.name = 'Calibri'
style.font.size = Pt(10.5)
style.font.color.rgb = RGBColor(0x33, 0x33, 0x33)
with open('tailored_company.md') as f:
lines = f.readlines()
for line in lines:
raw = line.strip()
if not raw:
continue
# Name line
if raw.isupper() and len(raw) < 50 and not ':' in raw:
p = doc.add_paragraph()
p.alignment = WD_ALIGN_PARAGRAPH.CENTER
r = p.add_run(raw)
r.bold = True
r.font.size = Pt(16)
continue
# Contact line
if '@' in raw or any(c.isdigit() for c in raw[:8]):
p = doc.add_paragraph()
p.alignment = WD_ALIGN_PARAGRAPH.CENTER
r = p.add_run(raw)
r.font.size = Pt(9.5)
r.font.color.rgb = RGBColor(0x55, 0x55, 0x55)
continue
# Section headers (all caps)
if raw.isupper() and not raw.startswith('-') and '|' not in raw:
p = doc.add_paragraph()
p.space_before = Pt(6)
r = p.add_run(raw)
r.bold = True
r.font.size = Pt(11)
continue
# Job lines with |
if '|' in raw and not raw.startswith('-') and not raw.startswith('โข'):
parts = [x.strip() for x in raw.split('|')]
p = doc.add_paragraph()
p.space_before = Pt(4)
p.space_after = Pt(0)
r = p.add_run(parts[0])
r.bold = True
r.font.size = Pt(10.5)
if len(parts) >= 2:
r2 = p.add_run(' | ' + parts[1])
r2.font.size = Pt(10.5)
r2.font.color.rgb = RGBColor(0x55, 0x55, 0x55)
if len(parts) >= 3:
p2 = doc.add_paragraph()
p2.space_before = Pt(0)
p2.space_after = Pt(1)
r3 = p2.add_run(parts[2].strip())
r3.font.size = Pt(9.5)
r3.font.color.rgb = RGBColor(0x66, 0x66, 0x66)
r3.italic = True
continue
# Bullets
if raw.startswith('- ') or raw.startswith('โข '):
text = raw.lstrip('-โข ').strip()
p = doc.add_paragraph(text, style='List Bullet')
p.space_before = Pt(0)
p.space_after = Pt(1)
for run in p.runs:
run.font.size = Pt(10)
continue
# Tech skill lines with colon
if ':' in raw and not raw.startswith('-'):
parts = raw.split(':', 1)
p = doc.add_paragraph()
p.space_before = Pt(1)
p.space_after = Pt(1)
r = p.add_run(parts[0] + ':')
r.bold = True
r.font.size = Pt(10)
if len(parts) > 1:
r2 = p.add_run(parts[1])
r2.font.size = Pt(10)
continue
# Regular text
p = doc.add_paragraph(raw)
for run in p.runs:
run.font.size = Pt(10)
for section in doc.sections:
section.top_margin = Inches(0.6)
section.bottom_margin = Inches(0.5)
section.left_margin = Inches(0.7)
section.right_margin = Inches(0.7)
output_path = os.path.expanduser('~/Documents/Resumes/Jason_Crabtree_resume_Company_Role.docx')
doc.save(output_path)
print(f"Saved to {output_path}")
After building the DOCX, verify that no markdown syntax leaked into the output before presenting it. The user will catch **, *, or [text](url) artifacts if they slip through.
Use the verification script from the skill:
source /tmp/.venv/bin/activate
python3 scripts/verify_docx.py ~/Documents/Resumes/tailored_company.docx
If artifacts are found, the script lists every paragraph containing them so you can debug the conversion. Common causes:
p.add_run(text) instead of _add_rich(p, text) โ **bold** markers pass through unprocessed- **Category:** items) matched the generic bullet handler before the skill handler โ reorder handlers so skill lines come first**Title** | Company was split on | but the ** in the first part wasn't strippedIf using the skill's scripts/md_to_docx.py (the canonical converter), this should never happen. If it does, file a bug or patch the converter.
# Quick inline check (no script needed):
source /tmp/.venv/bin/activate
python3 -c "
import docx
doc = docx.Document(os.path.expanduser('~/Documents/Resumes/tailored_company.docx'))
issues = [p.text for p in doc.paragraphs if '**' in p.text]
if issues:
print(f'โ ๏ธ Found {len(issues)} paragraphs with markdown artifacts:')
for t in issues[:5]:
print(f' {t[:80]}')
else:
print('โ No markdown artifacts found')
"
The Hermes system venv doesn't have python-docx. Use the same temp venv approach:
cd /tmp && uv venv .venv && source .venv/bin/activate && uv pip install python-docx
python3 -c "
import docx
doc = docx.Document(os.path.expanduser('~/Documents/Resumes/resume.docx'))
for p in doc.paragraphs:
print(p.text)
"
If /tmp/.venv already exists from a prior Stage 2 run, skip the uv venv step โ just activate and go.
When generating DOCX files, never use markdown syntax:
โ INCORRECT (markdown in DOCX):
**Professional Summary**
*Manager, Technical Support Engineering*
[LinkedIn](https://linkedin.com/in/...)
| Category | Tools |
|----------|-------|
โ CORRECT (Word native formatting):
run.bold = True for bold text, not **text**run.italic = True for italic text, not *text*Key Principle: DOCX files should contain only clean, formatted text with no markdown artifacts. The user may not use LinkedIn or want social media links on their resume.
Career Changers:
Recent Graduates:
Senior Executives:
Technical Roles:
Creative Roles:
This skill processes your personal and professional information to generate tailored resumes. Always review the output before submitting to ensure accuracy and appropriateness. Remove or modify any information you prefer not to share with potential employers.