Live — connect today
Lemmy has no central app to register. FounderReply logs in to whichever instance your account lives on, then drafts posts, comments and direct messages for your approval. Search reaches content federated from other instances, so one login covers conversations across much of the network rather than a single server.
On Lemmy
Instance login, federated search, comments and DMs — all human-approved.
Your instance already mirrors the communities it federates with, so searching from it reaches far past the server you signed up on.
Search runs against your instance’s own API, which means results are exactly the federated content your account can already see — no more, and no cross-instance credential.
Lemmy is one of the few platforms here where the API supports replying to other people and messaging them, so both are on the table — behind approval.
Commenting on others and private messages are both implemented, and both sit under the approval ceiling in the table below rather than under a feature flag.
The connect step logs in once and keeps the session token the instance hands back. The password is not retained after that exchange.
Login returns an instance JWT which becomes the stored credential; Lemmy issues long-lived tokens with no refresh endpoint, so there is nothing to re-authorise.
The room
Lemmy is where a chunk of the technical Reddit audience went after 2023, organised the same way: topic communities, threaded comments, and a strong bias toward people who are visibly part of the community. Instances are run by volunteers, so the tolerance for marketing is lower and the tolerance for a genuinely useful answer is higher.
Every instance writes its own rules and its admins can ban an account or defederate a whole server. That makes the downside of a bad automated comment much larger than on a centralised platform: you do not lose a post, you lose access to a community and everyone federating with it.
The ceiling
Two layers decide whether an action happens by itself: your autonomy dial, and a hard ceiling we impose per platform. The safer of the two always wins, so the dial can make things stricter and never looser. This table is generated from that ceiling.
| Action | Ships as | Most permissive allowed |
|---|---|---|
| Publish your own postOriginal content from your connected account | Waits for your approval | May run automatically |
| Reply on your own postsAnswering comments left on your content | Waits for your approval | May run automatically |
| Direct messagePrivate messages to a person | Waits for your approval | Waits for your approval |
| Comment on someone else’s postThe highest-risk action on every platform | Not permitted | Waits for your approval |
Lemmy instances ban spam and self-promotion at their own discretion and can defederate a server entirely. Publishing your own posts and replying on your own threads may run unattended; commenting on strangers and sending DMs are held for a human.
Lemmy has no central app registration and no OAuth. You supply an instance URL and your credentials once, and the instance JWT becomes the stored credential.
Nothing here is a marketing paraphrase — it is the string the connect flow sends, copied from app/api/v1/connect/lemmy.
POST /api/v1/connect/lemmy
{
"workspace_id": "ws_…",
"instance_url": "https://lemmy.world",
"username": "yourname",
"password": "…"
}FAQ
Any instance you already have an account on, given a reasonably current Lemmy version. There is no app to register and no per-instance approval, because Lemmy is open source and every server exposes the same API. The instance you connect is stored per account.
No. The connect step sends it once to your instance’s login endpoint and keeps only the session token that comes back. Lemmy tokens are long-lived with no refresh endpoint, so the password is never needed again and is not retained.
An instance admin can ban your account exactly as they could if you were typing by hand, and can defederate other servers. That is precisely why comments on other people’s threads are capped at human approval here — the cost of a bad automated comment on Lemmy is access, not a deleted post.
No. Lemmy is open source and its API is free with no keys and no paid tier, so it is available on every plan including Free. What it costs is judgement: instances are run by volunteers who are quicker to remove marketing than a platform team would be.
The free plan connects one account and drafts replies for you to approve. Nothing is published until you release it.