Resume Tailor Skill
Core Objectives
- •ATS Hit Rate: Cover high-frequency JD keywords while maintaining natural expression.
- •Job Relevance: Prioritize evidence chains and quantified results that best match the position.
- •Delivery Quality: Output A4 PDF with extractable text and pass format quality checks.
Inviolable Principles
- •No Fabrication: Only rewrite, rearrange, compress—no fictional experience/tech/data.
- •Review Before Export: Must provide complete review draft and obtain explicit approval before export.
- •Pass Volume Gate Before Review: Reduce content to target length first, then show full text to user.
- •One Question at a Time: Only one decision point per round, use
questiontool. - •ATS Friendly: No table-based main layout, no images replacing body text, no key info in headers/footers.
- •Fixed A4 + Single Page: Default delivery 210mm x 297mm, 1 page.
- •PDF Only Delivery: Final output is PDF only.
- •Call
pdfskill before PDF actions: Execute before initial generation, minor regeneration, or final export.
Stateless Boundary (Mandatory)
- •
resume-tailorskill directory can only store "rules + templates + scripts + references". - •No personalized cache (user preferences, historical JDs, contact info, resume samples) in skill directory.
- •Unified directories (relative to workspace):
- •
resume_output/: Latest PDF - •
resume_output/backup/: Historical PDF backups - •
cache/user-profile.md: Long-term preference cache - •
cache/resume-working.md: Current session resume body
- •
Minimum Execution Flow
- •Session Startup Cleanup:
scripts/resume_cache_manager.py reset - •Read Template Resume:
template-check→template-show(if not exists, thentemplate-init) - •JD Diagnosis: Output P1/P2/P3 matches and gaps, generate initial version after user confirmation
- •Iterative Updates: Only change 1 decision point per round,
updateafter confirmation - •Volume Gate: First reduce to single-page target, then output full review text
- •QA + Dehumanization: Call
humanizer, complete structure/quantification/ATS checks - •PDF Generation and QC: After user approval, call
pdfand generate, detect, re-adjust if needed - •Wrap-up: Update profile cache, keep working cache as baseline for future sessions
Complete checklist and thresholds in references/execution-checklist.md.
Fixed Template Baseline (Logical Structure)
Always organize content in the following module order:
markdown
Header Summary Technical Skills Professional Experience Education
Script Responsibility Boundaries
- •
scripts/resume_cache_manager.py: Managecache/resume-working.mdreset/init/update/show (cleanuponly manual as needed), andcache/base-resume.mdtemplate-init/template-use/template-show/template-check. - •
scripts/resume_md_to_json.py: Convert standardized Markdown to template input JSON. - •
scripts/generate_final_resume.py: Accept--input-mdor--input-jsonand generate final PDF. - •
scripts/check_pdf_quality.py: Perform general format and text quality checks. - •
templates/modern_resume_template.py: Only responsible for PDF layout and export.
Dependent Skills
- •
docx: Read.docxresumes - •
pdf: Read PDFs, execute all PDF generation and recheck actions - •
humanizer: Remove AI traces, enhance natural expression
Agent Installation and Auto-Execution Entry
- •Master installation entry:
docs/guide/installation.md - •Machine-readable manifest:
install/agent-install.yaml - •OpenCode command entry:
.opencode/command/install-skill-deps.md
Reference Materials
- •Full process checklist and thresholds:
references/execution-checklist.md - •ATS strategy:
references/ats-keywords-strategy.md - •Profile cache template:
references/profile-cache-template.md - •Working cache structure:
references/resume-working-schema.md - •Template instructions:
templates/README.md