Resume Intelligence

Resume Intelligence is a local Codex/Claude/Cursor-friendly plugin for building evidence-backed resumes, ATS-safe application files, recruiter-facing visual resumes, LinkedIn recommendations, job match reports, and interview prep from approved career evidence.

It is designed for profiles where evidence is spread across LinkedIn exports, local project documents, Confluence, Jira, public GitHub, GitHub Enterprise, open-source work, and job postings.

What It Produces

• source-inventory.md: every approved source, folder, connector scope, skipped file, and scan status.
• tool-usage-log.md: tools, commands, connector calls, searches, and manual inputs used.
• evidence.md: source-backed claims with confidence and disclosure rules.
• run-summary.md: concise audit of sources scanned, tools used, outputs created, quality warnings, sensitive approvals, and missing user answers.
• safe-local-mode-checklist.md: local-only run controls when connectors or uploads should stay disabled.
• disclosure-approval-queue.md and final-redaction-review.md: enterprise/private-source approval and final sharing checks.
• draft-resume.md and professional-resume.md: evidence-backed resume drafts.
• detailed-resume.md: full interview/master resume with timeline, roles, responsibilities, projects, impact, metrics, and missing-info questions.
• career-vault/ and career-vault-lint.md: Obsidian-friendly LLM wiki for reusable career knowledge and future optimization.
• career-master-run.md, missing-information-loop.md, and career-positioning-coach.md: career strategy, missing facts, and positioning guidance.
• ats-resume.docx: clean application resume for ATS/job portals.
• designed-resume.html and optional designed-resume.pdf: high-quality recruiter/referral visual resume.
• designed-resume.docx: optional editable Word companion, not the main visual surface.
• resume-scorecard.md: ATS, recruiter, page-count, keyword, and claim-risk review.
• targeted-resume.md, cover-letter.md, and gap-analysis.md: job-specific outputs.
• job-match-report.md: job search and match scoring.
• project-interview-briefs.md, technical-stack-interview-guide.md, interview-prep-pack.md, and interview-readiness-scorecard.md: interview prep and coaching.
• linkedin-recommendations.md: editable LinkedIn profile improvements.
• visual-template-selector.md, visual-design-review.md, and visual-tool-handoff.md: Figma/Canva/HTML/PDF design profile and handoff notes.

Repository Layout

.
|-- .agents/plugins/marketplace.json          # Codex local marketplace manifest
|-- .claude-plugin/marketplace.json           # Claude Code marketplace manifest
|-- plugins/resume-intelligence/
|   |-- .codex-plugin/plugin.json             # Codex plugin manifest
|   |-- .claude-plugin/plugin.json            # Claude Code plugin manifest
|   |-- .cursor-plugin/plugin.json            # Cursor plugin metadata
|   |-- skills/resume-intelligence/SKILL.md   # Main workflow
|   |-- skills/resume-intelligence/references/
|   |-- skills/resume-intelligence/assets/
|   |-- skills/resume-intelligence/assets/vault-templates/
|   |-- skills/resume-intelligence/scripts/
|   |-- skills/resume-intelligence/examples/
|   `-- validation/dry-run-scenarios.md
`-- docs/superpowers/

Install For Codex

From GitHub

After this repository is published:

codex plugin marketplace add rahulbsw/resume-builder
codex plugin add resume-intelligence@resume-builder-local

The Codex marketplace name is defined in .agents/plugins/marketplace.json as resume-builder-local.

From A Local Clone

git clone https://github.com/rahulbsw/resume-builder.git
cd resume-builder
codex plugin marketplace add "$(pwd)"
codex plugin add resume-intelligence@resume-builder-local

Update An Existing Local Codex Install

cd /path/to/resume-builder
python3 "$HOME/.codex/skills/.system/plugin-creator/scripts/update_plugin_cachebuster.py" plugins/resume-intelligence
codex plugin add resume-intelligence@resume-builder-local

Start a new Codex thread after reinstalling so new skill metadata and templates are loaded.

Validate Local Setup

From the repository root:

python3 plugins/resume-intelligence/skills/resume-intelligence/scripts/setup_doctor.py .

The setup doctor checks required plugin files and reports optional local tools such as rg, gh, codex, claude, cursor, npx, and soffice.

Install For Claude Code

This repo includes Claude Code plugin manifests.

Marketplace Install

claude plugin marketplace add rahulbsw/resume-builder
claude plugin install resume-intelligence@resume-builder

Local Install For Development

git clone https://github.com/rahulbsw/resume-builder.git
cd resume-builder
claude plugin marketplace add "$(pwd)"
claude plugin install resume-intelligence@resume-builder

One-Session Local Use

Use this when you want to test without installing a marketplace:

claude --plugin-dir plugins/resume-intelligence

Install For Cursor

Cursor support depends on the Cursor version and plugin/MCP features available in your environment. This repo includes .cursor-plugin/plugin.json metadata and can be used in two practical ways.

Project Context Use

git clone https://github.com/rahulbsw/resume-builder.git
cursor resume-builder

In Cursor, ask the agent to use:

Use the Resume Intelligence skill in plugins/resume-intelligence to build my ATS resume, visual recruiter resume, job match report, and interview prep.

MCP/Tool Setup

For best results in Cursor, add the same MCP tools you use in Codex or Claude: GitHub, Atlassian/Jira/Confluence, Figma, browser automation, and local filesystem access. Cursor supports adding MCP servers from the CLI:

cursor --add-mcp '{"name":"example-tool","command":"example-command","args":["--example"]}'

Use your real MCP server command and credentials. Do not commit tokens or secrets to this repository.

Install With skills.sh

Use this option when you want to install only the Resume Intelligence skill through the Skills CLI. For full Codex/Claude plugin metadata, marketplace install, and Codex cachebuster behavior, use the plugin install sections above.

You can run the CLI through npx:

npx skills add https://github.com/rahulbsw/resume-builder/tree/master/plugins/resume-intelligence/skills/resume-intelligence -a codex
npx skills add https://github.com/rahulbsw/resume-builder/tree/master/plugins/resume-intelligence/skills/resume-intelligence -a claude
npx skills add https://github.com/rahulbsw/resume-builder/tree/master/plugins/resume-intelligence/skills/resume-intelligence -a cursor

To install for every supported local agent:

npx skills add https://github.com/rahulbsw/resume-builder/tree/master/plugins/resume-intelligence/skills/resume-intelligence -g -a all

For local development from a clone:

git clone https://github.com/rahulbsw/resume-builder.git
cd resume-builder
npx skills add plugins/resume-intelligence/skills/resume-intelligence -a codex

If you prefer a global CLI install:

npm install -g @skills/cli
skills add https://github.com/rahulbsw/resume-builder/tree/master/plugins/resume-intelligence/skills/resume-intelligence -a codex

Optional Tools For Best Experience

Resume Intelligence works with pasted text and local files, but it is strongest when the agent has tools for source collection, document rendering, and visual design.

Recommended optional tools:

• GitHub connector or gh CLI: public GitHub, GitHub Enterprise, PRs, issues, commits, and open-source evidence.
• Confluence connector: design docs, launch notes, architecture docs, and project summaries.
• Jira connector: epics, tickets, incidents, delivery evidence, and project timelines.
• Browser or Chrome tool: review local HTML resumes and export PDF when supported.
• Documents/DOCX tooling: create ats-resume.docx and render-check Word output.
• LibreOffice soffice: required for reliable DOCX render QA in many local environments.
• skills.sh CLI: install the Resume Intelligence skill directly into Codex, Claude, Cursor, or all supported local agents.
• Obsidian: optional viewer/editor for career-vault/ graph view, backlinks, and manual review of the LLM wiki.
• Figma plugin/MCP: create or refine a high-quality visual resume when you approve a target Figma file.
• Canva: useful for manual visual polish; use the generated HTML/PDF and visual-tool-handoff.md as the handoff package.
• Web search: current job postings, market language, and job match scoring.

Useful Codex installs when available:

codex plugin add github@openai-curated
codex plugin add browser@openai-bundled
codex plugin add chrome@openai-bundled
codex plugin add documents@openai-primary-runtime
codex plugin add figma@claude-plugins-official

Some plugin names and marketplaces vary by Codex installation. Run this to inspect what your environment exposes:

codex plugin list

How To Use

Start with a new thread after installing or updating the plugin.

General Resume

Use $resume-intelligence to build my resume from:
- my LinkedIn export at <path or pasted text>
- local project docs under <folder>
- public GitHub user <username>
- GitHub Enterprise host <host> with internal-summary-only disclosure

Target role: Staff Platform Engineer
Years of experience: 12
Create ATS DOCX, professional HTML/PDF, LinkedIn recommendations, and interview prep.

Safe Local-Only Run

Use $resume-intelligence in safe local-only mode.
Use only my LinkedIn export or pasted profile, local project documents under <folder>, and local job posting text.
Create source inventory, tool usage log, evidence log, ATS resume, visual resume, career positioning coach, interview readiness scorecard, disclosure approval queue, and run summary.
Do not connect to external tools, browse job boards, or upload content to Canva/Figma unless I approve it later.

Job-Specific Resume And Cover Letter

Use $resume-intelligence to tailor my resume and cover letter for this job posting: <paste posting or URL>.
Consider my years of experience, target seniority, required technologies, and evidence-backed achievements.
Create gap analysis and keyword coverage without inventing unsupported claims.

Job Search And Match Scoring

Use $resume-intelligence to search for matching senior/staff platform engineering jobs in the United States.
Score each job against my resume evidence, years of experience, target technologies, and location preference.

Interview Prep

Use $resume-intelligence to prepare interview material from my strongest projects.
Create project deep dives, technical-stack preparation, behavioral stories, system design prompts, recruiter-screen talking points, and an interview-readiness scorecard.

Career And Interview Coach

Use $resume-intelligence as my career and interview coach.
Build a career master run, detailed interview resume, missing-information loop, career positioning coach, project briefs, technical-stack guide, interview prep pack, and interview readiness scorecard.
Use evidence-backed claims only and ask for missing dates, ownership, metrics, scale, and disclosure decisions.

Career Knowledge Vault / Obsidian LLM Wiki

Use $resume-intelligence to build a Career Knowledge Vault for future resume and interview optimization.
Create an Obsidian-friendly career-vault/ from my evidence with project pages, technology pages, disclosure notes, index.md, log.md, and AGENTS.md.
Run the vault linter and save career-vault-lint.md.
Keep raw sources separate and preserve source context references for every claim.

When structured evidence JSON is available, the helper can seed the vault:

python3 plugins/resume-intelligence/skills/resume-intelligence/scripts/build_career_vault.py resume-runs/<run>/evidence.json --output-dir resume-runs/<run>/career-vault --run-name "staff platform engineer" --run-date YYYY-MM-DD
python3 plugins/resume-intelligence/skills/resume-intelligence/scripts/vault_lint.py resume-runs/<run>/career-vault > resume-runs/<run>/career-vault-lint.md

Open resume-runs/<run>/career-vault/ as an Obsidian vault when you want graph view, backlinks, and manual review. Obsidian is optional; the vault is plain Markdown and works with Codex, Claude, Cursor, VS Code, or any editor.

Detailed Interview Resume / Master Timeline

Use $resume-intelligence to create a detailed interview resume.
Include my full work timeline, every role, responsibilities, important projects, ownership, architecture decisions, impact, metrics, technologies, and questions for any missing information.
Do not invent dates, metrics, or ownership. Ask me for anything that is missing or low-confidence.

Visual Resume

Use $resume-intelligence to create my ATS DOCX plus a high-quality recruiter HTML/PDF resume.
Use the visual template profile that best fits my target role: modern-technical, executive-compact, open-source-builder, enterprise-platform, or ai-automation.
Use visual-tool handoff if Canva or Figma would improve the design.
Do not upload private content to third-party tools without asking me first.

Create A Run Folder

python3 plugins/resume-intelligence/skills/resume-intelligence/scripts/resume_runner.py "staff platform engineer"

This creates resume-runs/YYYY-MM-DD-staff-platform-engineer/ and copies the standard Markdown templates for a new run.

Best Practices

• Keep ATS and visual outputs separate. Use ats-resume.docx for job portals and designed-resume.html or designed-resume.pdf for recruiter/referral sharing.
• Start with evidence. Require source-inventory.md, tool-usage-log.md, and evidence.md before polished resume wording.
• Keep current agent-guided behavior as the default. Helpers and quality gates improve visibility without blocking normal resume generation.
• Preserve public and enterprise GitHub as separate source contexts. They often have different identities, credentials, and disclosure rules.
• Treat enterprise Confluence, Jira, private repositories, customer names, internal URLs, and exact metrics as confidential until explicitly approved.
• Run setup_doctor.py after cloning or updating when install behavior looks inconsistent.
• Use career-master-run.md and interview-readiness-scorecard.md for preparation; use professional-resume.md or ats-resume.docx for actual applications.
• Use career-vault/ for durable knowledge across runs. Keep evidence.md as the provenance source and treat the vault as reusable synthesis.
• Do not rely on LinkedIn scraping. Use pasted LinkedIn content, exports, public profile notes, or user-provided summaries.
• Prefer high-confidence claims. Mark weak or inferred claims for user confirmation instead of turning them into polished bullets.
• Use live search only for current job matching. Job postings change quickly.
• Use exact job keywords only when the evidence supports them. Avoid hidden keyword stuffing.
• Prefer HTML/PDF/Figma for visual quality. DOCX is important for editability and ATS submission, but it is not the best high-fidelity visual surface.
• Review every generated output before applying. The plugin helps draft and organize; you remain responsible for final truthfulness and disclosure.

Optimization Tips

• Narrow connector queries by repo, project, space, date range, issue type, or keyword before pulling large histories.
• Use python3 plugins/resume-intelligence/skills/resume-intelligence/scripts/inventory_sources.py <approved-folder> --output resume-runs/<run>/source-inventory.md for a deterministic first-pass inventory of a local evidence folder.
• Keep local evidence folders organized by project or employer. Recursive inventory works best when folders have meaningful names.
• Convert important PDFs, DOCX files, and exported tickets to text-friendly formats when possible.
• Keep one run per target role or job posting under resume-runs/YYYY-MM-DD-<slug>/.
• Reuse confirmed evidence across runs instead of re-collecting everything from scratch.
• Focus interview prep on the top 3-5 projects that show architecture, tradeoffs, ownership, impact, and lessons learned.
• Use detailed-resume.md as the master source for interview preparation, then trim into professional-resume.md and targeted-resume.md for applications.
• Use Figma or Canva only after content is sanitized and stable; otherwise visual polish will amplify content churn.

Privacy And Credential Rules

• Never store credentials, tokens, cookies, or passwords in plugin files or generated outputs.
• Do not commit generated private resume runs unless you intentionally want them in Git.
• Do not upload private resume content, internal project details, or profile pictures to Canva, Figma, or another third-party service without explicit approval.
• Ask before including internal project names, customer names, repository names, Jira IDs, Confluence titles, private URLs, or exact internal metrics in final user-facing outputs.

Validate The Plugin

Codex validation:

python3 "$HOME/.codex/skills/.system/plugin-creator/scripts/validate_plugin.py" plugins/resume-intelligence

python3 "$HOME/.codex/skills/.system/skill-creator/scripts/quick_validate.py" plugins/resume-intelligence/skills/resume-intelligence

If your Codex validator environment requires a custom PYTHONPATH, reuse the value from your local Codex setup.

Claude validation:

claude plugin validate .
claude plugin validate plugins/resume-intelligence

Generated DOCX structural check:

unzip -t plugins/resume-intelligence/skills/resume-intelligence/assets/docx-templates/ats-safe-resume-template.docx
unzip -t plugins/resume-intelligence/skills/resume-intelligence/assets/docx-templates/modern-technical-resume-template.docx

Development Notes

After editing the Codex plugin, update the cachebuster and reinstall:

python3 "$HOME/.codex/skills/.system/plugin-creator/scripts/update_plugin_cachebuster.py" plugins/resume-intelligence
codex plugin add resume-intelligence@resume-builder-local

Then start a new Codex thread before testing behavior.