overleaf-resume
Enables local editing and compilation of LaTeX resumes, syncing with Overleaf via browser automation without Git or API access.
README
Overleaf Resume
Edit and compile your resume(s) locally instead of on the Overleaf website,
with an MCP server that syncs source files with Overleaf when needed. Works
with any number of Overleaf projects side by side — every sync tool takes an
explicit project_id, so nothing is pinned to a single "the" project.
Built for free-plan Overleaf accounts (no Git/API access). If your
account has Git integration (paid plan), just git clone your project
instead — you don't need this.
How it works
-
Day-to-day editing: once a project is pulled (see below), edit its
.texfile directly underprojects/<name>/, then compile locally with Tectonic — no Overleaf round trip needed:tools/tectonic.exe projects/<name>/<main file>.texThis produces a PDF right next to it. (
tools/tectonic.exeis a standalone binary already set up in this repo — gitignored since it's a large downloaded binary, not source.) -
Syncing with Overleaf: since this is a free-plan Overleaf account (no Git/API access), an MCP server (
overleaf_mcp/) drives a real, logged-in Chromium session with Playwright to download a project's source and push edits back into its online editor — the same actions you'd take by hand in the browser.
Project layout
Each Overleaf project you pull lands in its own subfolder, so multiple resumes can coexist without collisions:
projects/
<your-project-name>/
resume.tex (or whatever the project's main file is named)
...other pulled files (.cls, .sty, images)
.backups/ # timestamped backups made before each re-pull
<another-project-name>/
...
projects/ is gitignored — it holds your actual resume content (pulled
from Overleaf), so it stays local and never gets committed.
One-time setup
Steps 1-2 are already done in this repo (.venv/ has the dependencies
installed and Playwright's Chromium browser downloaded; tools/tectonic.exe
is ready to use). They're listed here in case you need to redo them on
another machine or after a clean checkout.
-
Create a virtual environment and install dependencies:
python -m venv .venv .venv\Scripts\activate pip install -r requirements.txt playwright install chromium -
Get Tectonic:
winget/chocomay need admin rights that aren't always available. The reliable no-admin option is to download the Windows binary directly and drop it intools/:gh release download tectonic@0.17.0 --repo tectonic-typesetting/tectonic \ --pattern "tectonic-0.17.0-x86_64-pc-windows-msvc.zip" --dir tools # then unzip it so tools/tectonic.exe existsVerify with
tools/tectonic.exe --version. -
Log in to Overleaf once so the MCP server can reuse the session:
.venv/Scripts/python.exe scripts/overleaf_login.pyA browser window opens to Overleaf's login page. Log in with email + password, not "Sign in with Google" — Google blocks its sign-in form inside any automated browser, so Google-only accounts need a password set on Overleaf first (Account Settings). The session is then saved to
.auth/storage_state.json(gitignored — never commit or share it). Re-run this script wheneveroverleaf_login_statusreports the session has expired. -
This repo's
.mcp.jsonregisters the server with Claude Code automatically. Restart Claude Code (or run/mcp) so it picks up theoverleaf-resumeserver. -
Call
overleaf_list_projectsto see all your Overleaf projects and their ids, thenoverleaf_pull(project_id)for whichever one(s) you want to work on. Files land underprojects/<name>/.
MCP tools
overleaf_login_status— check whether the saved session is still valid.overleaf_list_projects— list every Overleaf project (name + id).overleaf_pull(project_id, local_dir?)— download that project's source intoprojects/<name>/(orlocal_dirif given), backing up any existing.tex/.cls/.sty/.bibfiles there under.backups/first.overleaf_push(project_id, file_path)— push a local file's current content back into that Overleaf project's online editor.file_pathis relative to this repo root, e.g.projects/<your-project-name>/resume.tex.
project_id is always required — there's no single default project, so
this works the same whether you have one resume or several. Use
overleaf_list_projects whenever you need to look one up.
Troubleshooting local compiles
Overleaf compiles with pdfLaTeX by default; Tectonic uses a XeTeX-based engine. Two issues this surfaced on the popular "Jake's Resume"-style templates, in case you hit them too:
\pdfglyphtounicode/\input{glyphtounicode}/\pdfgentounicode=1are pdfTeX-only primitives. Guard them so both engines work:\usepackage{iftex} \ifPDFTeX \input{glyphtounicode} \fi ... \ifPDFTeX \pdfgentounicode=1 \fienumitem's\begin{itemize}...\end{itemize}pairs (via this template's\resumeSubHeadingListStart/Endmacros) must stay balanced even across commented-out sections — a stray, unmatched\end{itemize}left behind after commenting out a block produces a crypticUndefined control sequencedeep insideenumiteminternals (\enit@enditemize). This isn't a Tectonic/Overleaf difference — it'll fail on Overleaf's own compiler too if you look at the error count.
Caveats
- This automates Overleaf's actual web UI with your own logged-in
session — there's no public API for free-plan projects. It can break
if Overleaf changes its UI, and the saved session will occasionally
expire (re-run
scripts/overleaf_login.py). - The first
tectoniccompile needs network access to fetch LaTeX packages; later compiles are cached and work offline. .auth/storage_state.jsonholds live session cookies — keep it out of version control.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.