How to run your job search from Claude or ChatGPT

What changes when your assistant is connected

On its own, a chat assistant is a capable writer with no memory of your search. You paste a resume, paste a posting, get a draft, and the result lives in that one conversation. Nothing is tracked, nothing checks the draft against what you have actually done, and next week you start again.

Lamr runs an MCP server. MCP, the Model Context Protocol, is an open standard for connecting assistants to tools. Once Lamr is connected, the same assistant works on your actual workspace: your master resume, your applications, their status, notes and follow-up dates. When you ask for a tailored resume, Lamr's own tailoring runs with the same rules as the web app. It rewrites wording and order toward the posting, and it does not add numbers or experience your master resume does not contain.

The split of work is simple. Your assistant understands the request and picks the right step. Lamr does the tracking, scoring, tailoring and typesetting, and keeps a history of every change. You approve anything that spends credits and you do the applying.

Set it up once (about two minutes)

You need a Lamr account first. The free plan works. Then add Lamr as a connector using this server URL:

https://api.lamr.app/mcp

Claude (claude.ai or Claude Desktop): open Settings → Connectors, choose Add custom connector, name it Lamr, paste the URL and select Connect. Sign in to Lamr and approve access. If Claude does not reach for Lamr in a chat, check that the connector is switched on in that chat's tools menu.

ChatGPT: turn on developer mode, open connector settings, add a custom MCP connection with the same URL, and complete the Lamr sign-in. Whether developer mode and custom connectors are available depends on your ChatGPT plan, so check OpenAI's current help pages if you do not see the option.

Claude Code or Codex users can add it from the terminal:

claude mcp add --transport http lamr https://api.lamr.app/mcp

codex mcp add lamr --url https://api.lamr.app/mcp
codex mcp login lamr

You never paste a password or an API key into the chat. Sign-in uses OAuth, so the connection only ever acts as your own Lamr account, and you can remove it at any time from the same connector settings.

Check that it works

Start with something that reads and changes nothing. It costs no credits and confirms the connection is live:

Using Lamr, tell me how many applications I have in each status and how many credits I have left this cycle. Don't change anything.

If the very first reply takes up to a minute, that is Lamr waking up after a quiet period. Requests after that are quick.

The everyday loop, in four prompts

Most of a job search is the same four moves: decide whether a role is worth it, tailor for it, finish the documents, and record what you did. Each one is a single prompt.

1. Track it and check your fit. Screening scores you against every requirement in the posting before you spend anything on tailoring.

Track this job in Lamr and screen my fit. Use the full posting below, not a summary. Tell me my fit score, which requirements I don't meet, and whether it's worth tailoring for. Don't draft anything yet.

Link: [job posting URL, if you have one]

Posting:
[paste the full job posting]

2. Draft the documents once a role is worth it.

Draft my tailored resume and cover letter for the [Company] application in Lamr. Tell me what you changed and which requirements you couldn't support from my resume. Don't compile yet.

3. Finish and read them. Compiling is free, and so is switching templates later.

Compile the [Company] draft in Lamr with the classic template, then show me the final resume text here so I can read it.

4. Record what you did after you apply.

I applied to [Company] today. Mark it as applied in Lamr and set a follow-up date 10 days from now.

What it costs

Anything that only reads your workspace is free: listing applications, checking stats, reading a finished resume. AI work spends credits from the same monthly allowance as the web app, at the same price:

• Screen a posting: 1 credit.
• Draft a resume and cover letter: 5 credits, or 3 when an application only wants one of them.
• Interview prep: 2 credits.
• Answers to an employer's own questions: 3 credits, and 1 to redraft a single answer.
• Compile a PDF or switch its template: free.

The free plan includes 30 credits a month. Screening first is the habit that makes them last: screening ten postings costs 10 credits, while drafting for all ten would cost 50.

What it will not do

• Contact anyone. No emails, messages or replies go to recruiters or employers. Your assistant can draft a follow-up for you to send yourself.
• Apply on its own. Lamr has no browser. An assistant that controls a signed-in browser can use Lamr to fill in an employer's form, and Lamr refuses to answer self-identification and attestation questions you have not answered yourself. See how auto-apply works, and check the job site's terms before automating a submission.
• Invent experience. Tailoring does not add numbers that are not in your master resume, and your past job titles stay exactly as you wrote them. Requirements it cannot support show up as gaps.
• See payment details. The account tool returns your plan and credits left, nothing else.

Tools that spend credits, delete something or replace your resume are labelled that way, so your assistant asks you before it uses them.

Habits that get better results

• Name the application. Company is usually enough. If you have two roles at one company, add the role title.
• Paste the whole posting. Some job boards block automated fetching. If Lamr cannot import from a link, paste the full text instead.
• Say when you only want an answer. "Don't draft yet" and "don't change anything" keep a question from turning into a paid step.
• Ask for the history when something looks wrong. Lamr records every change and whether it came from the web app or your assistant.

Show me the change history for my [Company] application in Lamr, including whether each change came from the browser or from you.