ResuStack

ResuStack is an AI resume builder. Import a PDF or your LinkedIn profile, pick one of 14 designs, and edit by form or by chat — in the browser, or straight from Claude through its MCP server. Every change is reversible, and the PDF you download is rendered from the same template as the live preview.

🔗 Live: resustackapp.com

✨ Features

Resume builder

• PDF & LinkedIn import — upload an existing resume or a LinkedIn PDF; AI extracts and structures it
• Split-pane editor — the form on the left, a live preview on the right, with page breaks where the PDF will break
• 14 designs on 6 layouts — single column, banner, label gutter, header grid, left sidebar and right rail; ATS-safe designs are marked, and switching keeps your content
• English and Turkish resumes — headings, "Present", month names and degree phrasing print in the language the resume is written in
• "What I'm working on" — an optional section above Education, shown only when you tick it
• Sign in with Google — or with a username and password; email sign-ups confirm their address before using the AI features
• PDF export — rendered with WeasyPrint from the same template the preview uses

AI

• One-click enhance — rewrites experience and project descriptions into stronger bullet points
• Analyze and compare — score a resume and see where it is weak, or compare two versions
• Guided build — build a resume step by step through questions

Agentic mode

• Edit by chatting — "Make my last role sound more senior"; the agent uses tools, streams its progress and asks for approval before destructive actions
• Template pane — pick a design for the active resume without leaving the chat
• Undo — revert the last change from the conversation

Change history

• A restore point before every save and every AI or MCP edit
• Diff any version against the current one, and restore it — on every plan

Job applications

• Match a resume against a job posting and tailor a version for it
• Track applications, each with a snapshot of the exact resume you sent; clone a snapshot back into an editable resume

Language versions

• Translate a resume in place, or create a translated copy linked to the original

🤖 Use ResuStack from Claude (MCP)

ResuStack is a Model Context Protocol server. Claude — or any MCP client with Streamable HTTP and custom headers — writes the resume; ResuStack stores it, versions it and renders it.

1. Create a token. Sign in, open Profile → API token, and create one. It is shown once; replacing it revokes the old one. The Profile page also shows the endpoint address to use.

2. Add the server. With Claude Code:

claude mcp add --transport http resustack https://resustackapp.com/mcp --header "Authorization: Bearer YOUR_TOKEN"

Authentication is a bearer token only; session cookies are not accepted on this endpoint.

3. Ask. "List my resumes", "Switch my CV to the Label Gutter design and give me the PDF", "Fill What I'm working on from what we did this month."

Tools

Prompts

Safety rules

• Every write takes a restore point first; you can undo it on the website
• There is no delete tool — removing a resume stays on the website, where a person clicks
• Every query is scoped to the token's owner
• Rate limited per account

Listing in the MCP Registry

server.json at the repository root describes the server for the official MCP Registry as com.resustackapp/resustack. Publishing under that name requires proving ownership of resustackapp.com with a file served at /.well-known/mcp-registry-auth:

1. Generate a key pair locally. Never commit key.pem (it is in .gitignore):

   bash

   Copy

   openssl genpkey -algorithm Ed25519 -out key.pem
   

2. Print the proof record and set it as the MCP_REGISTRY_AUTH environment variable in Dokploy, then redeploy:

   bash

   Copy

   echo "v=MCPv1; k=ed25519; p=$(openssl pkey -in key.pem -pubout -outform DER | tail -c 32 | base64)"
   

3. Check it is live:

   Terminal

   Copy

   curl https://resustackapp.com/.well-known/mcp-registry-auth
   

4. Log in and publish from the repository root:

   bash

   Copy

   mcp-publisher login http --domain resustackapp.com --private-key "$(openssl pkey -in key.pem -noout -text | grep -A3 'priv:' | tail -n +2 | tr -d ' :\n')"
   

   bash

   Copy

   mcp-publisher publish
   

Bump version in server.json (and SERVER_INFO in mcp_server/protocol.py) for each new listing.

📸 Screenshots

Editor — form, live preview and the template pane

Choosing a design

Agentic mode

Dashboard

💳 Plans

Free to start. Pro is a one-time purchase for a period — no subscription.

Pro removes these limits. Current prices are on the pricing page. Limits live in FREE_TIER_LIMITS in core/settings.py.

🚀 Running locally

Prerequisites

• Docker & Docker Compose
• An OpenAI API key

1. Clone and configure

git clone https://github.com/koksalkapucuoglu/resume-enhance.git

cd resume-enhance && cp .env.example .env

2. Start

docker compose up --build

The app runs at http://localhost:8000.

3. Run the tests

docker compose exec web python manage.py test resume mcp_server

WeasyPrint and OpenAI are mocked in the unit tests; no API calls are made.

☁️ Deployment

Production runs on Dokploy. Every push to main triggers a deploy through a GitHub webhook: Dokploy builds the Dockerfile, and entrypoint.sh runs migrate and collectstatic before starting Gunicorn. Traefik handles HTTPS.

Setting it up on a new server:

1. Install Dokploy: curl -sSL https://dokploy.com/install.sh | sh
2. Open http://YOUR_SERVER_IP:3000 and create an admin account
3. Create a project → add an Application → connect this GitHub repository
4. Add a PostgreSQL service in the same project
5. Set the environment variables (see .env.prod.example) and the domain, then deploy

Notes:

• Leave Dokploy's Run Command empty — the Dockerfile's ENTRYPOINT does everything
• Behind Cloudflare's proxy, set Dokploy's domain encryption to None and Cloudflare SSL to Full
• The image installs the fonts the resume designs use; nothing is fetched at render time

docker-compose.prod.yml and the Caddyfile are kept for self-hosting without Dokploy; they are not what production uses.

🔑 Google sign-in

Sign-in with Google uses django-allauth. ResuStack's own login, sign-up and password pages stay in charge; allauth adds only the Google flow.