Use Coverlttr From Any AI Tool - Claude, ChatGPT, Cursor, and More

Coverlttr exposes an MCP server - a standard interface that lets any compatible AI tool call into your account directly. That means you can generate a tailored cover letter, apply a PDF template, and grab a download link, all from a conversation in Claude, ChatGPT, Cursor, or Gemini, without opening the Coverlttr web interface.

The install takes about thirty seconds and uses OAuth - no API keys to copy, no JSON to edit unless you want to.

Connect Your AI Tool

Pick your platform and follow the prompt. Most have a one-click install link that pre-fills everything.

After connecting, sign in to your Coverlttr account when the browser tab opens. That’s it - the tools are live in your next conversation.

What MCP Is (The Short Version)

MCP (Model Context Protocol) is an open standard that lets AI assistants call external tools. Instead of pasting your CV and job description into a chat and hoping the model figures it out, you give the model a structured tool that already knows how to do the job - with access to your uploaded CVs, saved listings, and generation engine.

The same MCP server works across every supported platform. You connect once, and it’s available everywhere you use that tool.

What You Can Do

Once connected, your AI assistant can access the following Coverlttr tools in natural conversation:

Check your account

• get_user_profile - see your name and current credit balance before committing to a generation
• get_cover_letter_cost - get a credit estimate for a specific job description before generating

Browse your saved content

• list_user_files - see the CVs and resumes you’ve uploaded to Coverlttr
• list_user_job_listings / get_user_job_listing - browse and retrieve your saved job listings, including the full stored job description

Save and generate

• save_user_job_listing - save a new job listing (company, role, job description) directly from the conversation
• create_cover_letter - generate a tailored cover letter from a saved listing and one of your uploaded CVs
• regenerate_cover_letter - produce a revised version of an existing letter with updated instructions; the original is preserved

Format and export

• list_templates - browse available LaTeX cover letter templates
• apply_template - render your letter into a chosen template and compile it to PDF
• export_file - export in PDF, DOCX, Markdown, LaTeX, or PNG
• get_download_url - get a short-lived download link for any exported file

You don’t need to call these explicitly. Just describe what you want and your AI tool will figure out which tools to use.

A Typical Workflow

Here’s what a real session looks like in Claude or ChatGPT after connecting:

You: I want to apply for a Senior Product Manager role at Notion. Here’s the job description: [paste]. Save it and write me a cover letter using my most recent CV.

The assistant calls save_user_job_listing to store the listing, list_user_files to find your CV, get_cover_letter_cost to confirm the credit cost, then create_cover_letter to generate the letter. It presents the result and tells you how many credits were used.

You: Apply the Modern Professional template and export it as a PDF.

The assistant calls list_templates, apply_template, export_file, then get_download_url - and hands you a link to download the finished PDF.

The whole thing takes under a minute, from paste to download link.

Authentication

Everything uses OAuth - your AI tool redirects you to Coverlttr to sign in, and the connection is authorized at that point. There are no API keys to generate, store, or rotate. Revoking access is done from your Coverlttr account settings.

Note: Uploading a CV isn’t supported through MCP - that’s done once from the Coverlttr settings page. Once your CV is uploaded, MCP tools can reference it in any conversation.

If You’re a Developer

The MCP server is a standard remote HTTP server. If you want to call it programmatically from a script or build tooling around it, point any MCP client SDK at the server URL (shown in the install widget above) and complete the OAuth flow once. From there you have full access to the tool catalog via the standard MCP protocol.

---

If you build something interesting on top - a job board integration, a batch-processing script, a Notion workflow - we’d love to hear about it.