Does lemonly have an MCP server?
Yes. lemonly is an AI-native applicant tracking and screening app with a hosted, remote MCP (Model Context Protocol) server at https://app.lemonly.ai/api/mcp. It works with Claude (web and desktop), ChatGPT, Claude Code and the Codex CLI over Streamable HTTP, and signs in with OAuth, so there is no API key. It offers 21 tools: your AI can list roles, read candidates, their per-requirement assessments, résumés and interview answers, see the pipeline, move, favorite, reject and restore candidates, comment, edit shared notes, edit role details, interview questions and requirements, and add, rename, reorder or delete empty pipeline stages. It follows the dashboard’s rules: no admin access, hidden roles stay hidden, contact details only from the Reached out stage, and every change is recorded as the user. The MCP server and lemonly’s AI candidate screening are free, with no credit card; sign up at https://app.lemonly.ai. Last updated October 22, 2026.
The AI-native ATS
The ATS with a built-in MCP server
Connect Claude, ChatGPT, Claude Code or Codex to lemonly and let your AI review, move and reject candidates for you – free, with no API key.
Last updated October 22, 2026
The short answer
Yes – lemonly has a hosted, remote MCP server at https://app.lemonly.ai/api/mcp. You sign in with OAuth and never handle an API key. Your AI gets the same access you have in the dashboard, under the same rules, and every change it makes is recorded as you.
21
tools for your AI
$0
screening and MCP
0
API keys to manage
In your AI app
What you can ask your AI to do
Hiring work moves into the chat you already have open. Ask in plain language; your AI calls lemonly’s tools and answers with real data from your pipeline.
“Which new candidates for the Product Designer role should I look at first?”
Ranks unopened candidates by fit
“Summarize Maya’s résumé and how she scored on each requirement.”
Reads the résumé and the per-requirement assessment
“Move Maya to Phone screen and note that she’s free on Thursdays.”
Moves the stage and edits shared notes
“Reject everyone who failed the location requirement, with a short reason.”
Filters by assessment, rejects with a reason
“Add a Take-home stage after Phone screen.”
Adds a pipeline stage in place
“Rewrite the interview questions to focus on async communication.”
Replaces the role’s interview questions
What it can do
Your AI acts as you, under the dashboard’s own rules. Reading a candidate never marks them as viewed, so new candidates stay new until you open them yourself.
It can read
- Your roles, their requirements and pipeline stages
- Candidates, their assessments, résumés and attachments
- Interview answers, shared notes and comments
It can change
- Move, favorite, reject or restore candidates
- Comment and edit shared notes, as you
- Edit role details, interview questions and requirements
- Add, rename, reorder and delete empty stages
Two-minute setup
How to connect lemonly to Claude or ChatGPT
First, sign up free at app.lemonly.ai. Then add the server URL to your AI app. When a lemonly page opens, sign in and click Allow access. If lemonly emails you a sign-in link, open it on the same computer.
Server URL: https://app.lemonly.ai/api/mcp. In lemonly, Settings → AI apps → Copy setup prompt gives you a prompt that sets it up for you.
Claude (web or desktop)
- 1Settings → Connectors → Add custom connector
- 2Paste the server URL and click Connect
ChatGPT
- 1Settings → Connectors (turn on developer mode if asked) → Create
- 2Paste the server URL and choose OAuth
Claude Code
claude mcp add --transport http lemonly https://app.lemonly.ai/api/mcp
- 1Run the command, then /mcp and choose lemonly
Codex CLI
[mcp_servers.lemonly] url = "https://app.lemonly.ai/api/mcp"
- 1Add this to ~/.codex/config.toml
- 2Run codex mcp login lemonly
What “AI-native ATS” means at lemonly
Most applicant tracking systems were built for people clicking through screens, with AI added later. lemonly was built the other way round: AI does the first read on every applicant, and your own AI is a first-class user of the same product.
AI does the first read
Every applicant is assessed against each of your requirements, and a chat interview asks your questions. Change a requirement and candidates are re-scored.
Your AI is a teammate
21 tools cover what the dashboard does day to day, with the same permissions, the same audit trail and the same Slack notices.
People stay in charge
Your AI proposes and acts only when you ask. Contact details stay gated, hidden roles stay hidden, and you can disconnect an app any time.
MCP server vs ATS API vs CSV export
Three ways to get hiring data in front of an AI. An MCP server is the only one a non-engineer can set up in minutes and that stays live and able to act.
| MCP server | ATS API | CSV export | |
|---|---|---|---|
| Setup | Paste one URL | Write and host code | Export, then upload |
| Sign-in | OAuth – no API key | An API key to guard | None – the file travels |
| Who can use it | Anyone on the hiring team | Engineers | Whoever has the file |
| Data | Live on every question | Live, once built | Stale once exported |
| Can take action | Yes – dashboard rules apply | Yes, once built | No |
| Best for | Day-to-day hiring in your AI app | Custom integrations | One-off analysis |
Security in short: OAuth sign-in with a consent screen that lists what the app can do, tokens that only work on the MCP server, per-user access, rate limits on comments and requirement edits, and one-click Disconnect under Settings → AI apps.
Keep reading
More on lemonly screening
Ready to try it? Sign up free at app.lemonly.ai, then connect your AI app in Settings → AI apps.
MCP server questions
An MCP server is a connector that lets an AI app use another product. MCP, the Model Context Protocol, is an open standard Anthropic introduced in 2024 and that ChatGPT, Claude and Codex now support. An ATS with an MCP server lets your AI assistant read candidates and act on your pipeline directly, instead of you copying résumés into a chat.
Yes. lemonly runs a hosted, remote MCP server at https://app.lemonly.ai/api/mcp. Add it to Claude, ChatGPT, Claude Code or Codex, sign in to lemonly and click Allow access. Your AI can then read your roles and candidates, move people through your pipeline, comment and edit roles, with the same access you have in the lemonly dashboard.
Claude on the web and desktop, ChatGPT, Claude Code and the Codex CLI all work today. Any AI app that supports remote MCP servers over Streamable HTTP with OAuth sign-in should connect the same way: paste the server URL, sign in to lemonly and approve access.
No. The lemonly MCP server signs you in with OAuth, the same sign-in you use for the dashboard, so there is no key to create, copy or rotate. To cut an AI app off later, open Settings in lemonly, find AI apps and click Disconnect.
Only from the Reached out stage onward, the same rule the lemonly dashboard follows. Before that, your AI sees a candidate’s assessment, résumé, career history and interview answers, but not their email or phone number. Candidate search never matches on email or phone either.
It can change it, within limits. Your AI can move, favorite, reject or restore candidates, comment, edit shared notes, update role details, interview questions and requirements, and manage pipeline stages. It cannot delete candidates or roles, delete a stage that still has candidates, or reach admin settings, and every change is recorded as you.
No. The MCP server is free, and so is lemonly’s AI candidate screening: no credit card, no contract and no per-role fee. The only paid lemonly product is sourcing, where lemonly finds candidates for you for a flat $8,000 per hire, due only when the hire starts.
Yes. lemonly imports applicants from Ashby and screens them against your requirements, and your AI then works with those candidates through the MCP server like any others. Connecting Ashby takes a few minutes with an Ashby API key, and the lemonly Ashby setup guide walks through each step.
Give your AI a seat on the hiring team.
lemonly screening and the MCP server are free. No credit card, no API key.