Have your agent call mine.
⭐ Use this as a template — fork it, delete the examples, then let an agent help you populate a complete Career OS.
Engagement Stack is a human-first, agent-readable Career OS: a source-controlled professional dossier for CV, work history, portfolio evidence, engagement preferences, structured offer requests, evaluation rubrics, trade-off rules, and job-search submission materials.
It is not just a resume template. It is a framework for maintaining a complete professional record that humans can browse directly on GitHub and agents can read, cite, tailor, and transform without inventing facts.
- 🚀 Start Here
- 🎯 What This Repo Is For
- 🧠 Core Rule
- 🙏 Inspiration
- 🔒 Public Repo, Private Overlay
- 📦 Repository Layout
- 🧪 Examples
- 📄 Resume And Submission Outputs
- 🌐 Human-Readable Landing Page
- 🎛️ Resume Shaders
- 🖼️ Brand Assets
- 🔗 Built With Engagement Stack
- Set Up Your Stack
- Agent Instructions
- Profile
- CV And Work History
- Portfolio
- Evidence
- Engagement Preferences
- Evaluation Rubric
- Submission Adapters
- Maintenance System
Engagement Stack answers five practical questions:
- Who is this person? See
profile/. - What have they done? See
cv/andportfolio/. - What proves it? See
evidence/and each project claim map. - What engagement do they want now? See
engagements/. - How should offers be evaluated? See
engagements/evaluation-rubric.md. - What can be generated for a channel? See
submissions/andgenerated/. - How should a resume be specialized? See
resume-shaders/.
The root README and folder READMEs are the default human-readable website. No publishing step is required.
Markdown is the human source of truth. YAML is the agent and automation surface.
Every important human document can have a YAML sidecar. The Markdown explains the person and their work. The YAML gives agents structured fields for routing, matching, scoring, validation, and export.
Engagement Stack is designed to be useful as a public repo without exposing private negotiation data.
Keep public-safe priorities in engagements/, such as preferred engagement models, remote/hybrid/in-office posture, offer packet requirements, and general trade-off rules. Keep sensitive data in a local private/ overlay, such as salary floors, target compensation, immigration details, current-employer constraints, family or health constraints, and negotiation thresholds.
The repo includes private.example/ as a copyable starting point. Copy it to private/ and fill real values there. .gitignore keeps private/ and *.private.* files out of version control by default.
Agents should read public files first, then private overlays if present. If private overlays are absent, agents must not infer hidden salary ranges or constraints; they should ask.
Engagement Stack is inspired in part by Every Inc.'s after-automation-agent-mode, especially the idea that a repository can become a durable working surface for agents: local instructions, source material, prompts, review criteria, and outputs arranged so humans and agents can collaborate from the same project context.
This project applies that pattern to a different domain: a professional Career OS for CV, portfolio evidence, engagement preferences, offer handling, and job-search submission materials.
.
├── README.md
├── AGENTS.md
├── setup/ # onboarding and completeness checks
├── framework/ # framework origin, version, adoption, link-back
├── profile/ # identity, positioning, operating style
├── cv/ # canonical CV, work history, skills, resume source
├── portfolio/ # projects, claims, case studies, proof
├── evidence/ # proof index: repos, products, artifacts, references
├── career-strategy/ # target roles, markets, positioning, search strategy
├── engagements/ # desired engagements and request instructions
├── resume-shaders/ # role/industry-specific resume shaping rules
├── resumes/ # short/long variants, manifest, preserved originals
├── styles/ # presentation themes (navy default, density-aware)
├── submissions/ # LinkedIn, ATS, USAJOBS, Indeed, email adapters
├── prompts/ # paste-ready agent prompts
├── integrations/ # optional upstream/downstream repo contracts
├── maintenance/ # refresh cadence and audit checklists
├── private.example/ # copy to ignored private/ for sensitive overlays
├── templates/ # blank files to copy into a user fork
├── examples/ # disposable fictional example content
├── scripts/ # optional local convenience scripts
└── generated/ # generated artifacts; safe to rebuild
The examples/fictional-engineer/ folder is a disposable fixture. It exists to show a complete shape. Delete it when adopting this framework, or keep it as reference. Anything with .example.md or .example.yaml is sample content, not a user fact.
Two canonical resume lengths, both navy-themed, plus a matching cover letter and ATS-plain exports. scripts/render-resumes.sh [short|long|cover-letter|ats|all] produces:
- short resume PDF — ~2 pages, tight density (the default submission resume)
- long resume PDF — ~3 pages, roomy density
- ATS DOCX — plain, derived from the short variant
- ATS TXT — plain, derived from the short variant
- cover-letter PDF — ~1 page, letter density, matching navy theme
The designed PDFs are rendered Pandoc → HTML → styles/navy.css → headless Chrome (Chromium in CI); the ATS DOCX/TXT come straight from Pandoc (intentionally plain). Outputs land in generated/resumes/ (gitignored). See resumes/ for the variants and manifest.yaml control surface.
No local setup required: the Render Resumes GitHub Actions workflow installs Pandoc + Chromium and runs the script for full parity. (USAJOBS/federal-style output is a separate future concern, not yet implemented.)
Ground-truth first: if you already have polished resumes, Engagement Stack uses them as-is — drop them in resumes/short.md / resumes/long.md, preserve originals verbatim in resumes/originals/, and skip resume generation unless you ask to refine. Shaders still layer on top for role/industry specifics.
Other submission materials — LinkedIn profile text, generic ATS field packs, and engagement-request replies — are generated from submissions/ and cv/.
The GitHub-rendered README is the default human surface. For a higher-polish option, you can also generate a designed landing page — a self-contained, navy-themed docs/index.html that fronts the same canonical facts: hero, signature-stat band, experience timeline, skills, open-source proof, and availability, with prominent Résumé and 🤖 Agents call-to-action links. The Agents button points at AGENTS.md, so a visitor's agent gets a one-click door to your machine-readable surface.
- Prompt:
prompts/generate-landing-page.md - Template:
submissions/web/landing-page.template.html— the proven page, genericized to{{PLACEHOLDER}}tokens - Adapter:
submissions/web/README.md
Run the prompt against a populated stack: it fills the template from profile/, cv/, evidence/, and engagements/ (public-safe — never compensation or private-overlay data), writes docs/index.html + docs/.nojekyll, and deploys via GitHub Pages (/docs on a public repo, or a <username>.github.io user-site). Live reference instance: https://mikebengtson.github.io/.
Most people keep a core resume and tailor it for a role, industry, or client. Engagement Stack makes that explicit.
A resume shader is a small Markdown/YAML pair that tells an agent how to shape the canonical CV before emitting a scoped resume. It can prioritize certain projects, suppress irrelevant details, and adjust language for a target industry without changing the canonical CV.
Shaders are the content-emphasis axis — what to emphasize for a target. They are orthogonal to format (how long, how it looks), which is handled by resumes/ + styles/: the short (~2pg) and long (~3pg) variants and the navy theme. Supplied ground-truth resumes are used as-is, with shaders layered on top to emit tailored copies.
Public-safe shaders can live in resume-shaders/. Sensitive or highly targeted shaded resumes can live in a private overlay or in generated/.
This README banner and the social preview image live in resources/branding/.
- README banner: resources/branding/banner/engagement-stack-banner.png
- Social preview: resources/branding/social/engagement-stack-social.png
Use the social preview in GitHub repository settings or on LinkedIn/X posts introducing the framework.
This repository is the framework source. A user fork should keep a link-back so agents and humans can understand the structure and find updates.
Framework docs: framework/origin.md