Introduction
What the done. API does, how to start, and what it never does.
This page says what the done. API does, how the guides fit together, and what it will never do for you.
The API lives at https://api.done.com/v1. Every call is plain HTTPS with JSON bodies. The API reference is generated from the API's own description, so it always matches what the API does.
Preview. The done. API opens soon. Nothing here is live yet, and the addresses and examples describe how it will work at launch.
What the API covers
The API covers everything done. does, from one small job to running outreach on autopilot.
- Building blocks. Score a person or a list against your customer profile, research a company, look people up, find a work email, and write one message in your voice.
- Scouts. Keep finding people who fit, on a schedule, and hand the good ones to a list or a campaign.
- The full workflow. Build, launch and run campaigns, read and answer replies, approve drafts, and hear about everything by webhook.
- Goals. Describe who you want to meet and how many meetings a week you want. done. proposes the setup, runs it after your yes, and reports back.
Three ways to use it
You do not need all of it. Most teams start at the first level and move up.
- Results only. Ask for scores, researched customer profiles, scouts, work emails and written messages. This is the API-only plan. It sells results, never raw LinkedIn data.
- The full workflow, driven by you. Everything above, plus people search and lookup, campaigns, sending accounts, drafts, conversations and replies.
- Goals on autopilot. The full workflow, with goals on top. You say what you want. done. proposes a plan, waits for your approval, then runs it.
Operations that are not part of the API-only plan say so in the API reference.
Start here
- Make a key in done. under Settings, API keys.
- Call
GET /meto see what the key can do. - Follow the Quickstart to score your first person.
curl https://api.done.com/v1/me -H "Authorization: Bearer $DONE_API_KEY"Keys that start with done_test_ work only in your organization's test workspace. Nothing a test key creates can ever send for real, and test work charges no credits. Read more in Keys, workspaces and sign-in.
How every endpoint behaves
These rules hold everywhere. Each has its own guide.
| Topic | In short | Guide |
|---|---|---|
| Keys and workspaces | A key has a reach, a permission, a daily credit cap and an expiry. | Keys, workspaces and sign-in |
| Versions | Your key is pinned to the API version of the day it was made. | Versions |
| Errors | One shape for every error, with a stable code and a fix. | Errors |
| Retries | Creates, paid work and sends need an Idempotency-Key. | Safe retries |
| Lists | limit, cursor, has_more and next_cursor. | Lists and pages |
| Limits | 120 calls a minute per key. | Rate limits |
| Costs | 1 credit is scoring one person. | Credits and costs |
| Long work | Scoring a list, a scout run and company research finish by webhook. | Webhooks |
Every answer, success or error, carries a Request-Id header. Quote it when you contact support.
What done. never does through the API
- It never sends to LinkedIn outside a campaign or a reply.
- It never skips an account's pacing, limits or sending hours.
- It never automates replies to people who raised their hand. You hear about them by the
person.repliedevent and answer them yourself. - It never shows how a score is worked out. A score comes with its fit, its label and one sentence of reason.
Sends made through sign-in, which means an AI assistant or the MCP server, are recorded as API sends. They are never recorded as the signed-in person's own reply, so sending hours and limits always apply.
Where to go next
- Quickstart: make a key and score a person.
- Building blocks: customer profiles, scoring, people and lists, and writing.
- Webhooks: receive and verify events.
- AI assistants and MCP: use done. from an assistant, with sign-in instead of keys.