Prefer it still? Press here
From your first reply to the bot of your dreams
A Fluxer-native bot SDK for JavaScript, TypeScript and Effect that starts in one file and gives you full control when you need it
This is the whole bot, one file. Send !ping and watch the handler answer
This chat is a preview. The test that fluxerly init writes sends !ping and hello to the SDK's in-memory Fluxer and checks the answers
import { runBot } from "@neontechspace/fluxerly"// A failure that stops the bot, such as a rejected token, is logged and sets a failing exit codeawait runBot({ token: processnpx @neontechspace/node --env-file=.env bot.jsI write software both as my job and for the love of it. I had wanted to build my own bot SDK for a while, and at some point I stopped playing with the thought and started doing it
Over the years, software I used and libraries I built on kept changing under me. Sooner or later an update lowered the quality or the work slowed down, and I paid for it along with everyone else who used them
I wanted something for myself that I know I can rely on. With the knowledge to build it, it felt like the right moment to start realizing my dreams
Since AI has made code cheap and I strive for quality, I chose to make it usable for everybody. So Fluxerly is a Fluxer-native SDK instead of a Discord-compatible layer. It follows modern developer standards and is made for developers of every size, whether you code with or without AI
| Day | Downloads | Running total |
|---|---|---|
| Sep 14 | 52 | 52 |
| Sep 15 | 0 | 52 |
| Sep 16 | 211 | 263 |
| Sep 17 | 19 | 282 |
| Sep 18 | 7 | 289 |
| Sep 19 | 3 | 292 |
| Sep 20 | 9 | 301 |
| Sep 21 | 2 | 303 |
| Sep 22 | 1 | 304 |
| Sep 23 | 160 | 464 |
| Sep 24 | 24 | 488 |
| Sep 25 | 17 | 505 |
| Sep 26 | 12 | 517 |
| Sep 27 | 380 | 897 |
| Sep 28 | 45 | 942 |
| Sep 29 | 1 | 943 |
| Sep 30 | 184 | 1,127 |
| Oct 1 | 31 | 1,158 |
| Oct 2 | 16 | 1,174 |
| Oct 3 | 211 | 1,385 |
| Oct 4 | 20 | 1,405 |
| Oct 5 | 12 | 1,417 |
| Oct 6 | 0 | 1,417 |
| Oct 7 | 69 | 1,486 |
| Oct 8 | 15 | 1,501 |
| Oct 9 | 168 | 1,669 |
| Oct 10 | 41 | 1,710 |
| Oct 11 | 21 | 1,731 |
| Oct 12 | 8 | 1,739 |
| Oct 13 | 0 | 1,739 |
| Oct 14 | 0 | 1,739 |
| Oct 15 | 0 | 1,739 |
| Oct 16 | 8 | 1,747 |
| Oct 17 | 10 | 1,757 |
| Oct 18 | 488 | 2,245 |
| Oct 19 | 0 | 2,245 |
| Oct 20 | 0 | 2,245 |
| Oct 21 | 0 | 2,245 |
| Oct 22 | 48 | 2,293 |
| Oct 23 | 259 | 2,552 |
| Oct 24 | 243 | 2,795 |
| Oct 25 | 35 | 2,830 |
| Oct 26 | 24 | 2,854 |
| Oct 27 | 23 | 2,877 |
| Oct 28 | 47 | 2,924 |
| Oct 29 | 20 | 2,944 |
Six parts that come with the SDK, each with the guide that explains it
Network calls return a Result, a success or a failure, instead of throwing, so your code handles expected errors where they happen. The SDK masks the bot token in everything it logs
if (sent.isErr()) throw sent.error 
Prefix commands such as !greet come with typed arguments, guards, cooldowns and groups. Help is generated from the same definitions
arguments: { name: { type: "text", rest: true } },cooldown: { durationMs: 5_000, per: "user" },
createTestBot runs the bot's own options against an in-memory Fluxer, so tests need no network or account
const replies = await bot.say("!ping") 
When Fluxer answers 429, the SDK waits and retries within the request's deadline. A write with an uncertain outcome is never repeated automatically
WARNratelimit[ratelimit.wait]
Ctrl+C gives running handlers time to finish, then stops cleanly. A bot that saves its sessions at that stop can pick up missed events if it restarts within a minute, and the optional supervisor restarts crashed processes
INFOlifecycle[lifecycle.shutdownComplete]
Automatic sharding splits the connection to Fluxer, one shard per 2,000 communities. The observe option reports requests and handlers for metrics and traces
sharding: "auto",The same SDK at three sizes. Each step keeps everything from the one before it
One options object in a plain JavaScript file, which Node.js runs without a build step
import { runBot } from "@neontechspace/fluxerly"await runBot({ token: process.env.FLUXER_BOT_TOKEN, events: { messageCreate: ({ message, reply }) => { if (message.content !== "!ping") return return reply("Pong!") }, },})Own the connection and the shutdown when you need to. createClient leaves connecting, running and stopping to your code
import { createClient } from "@neontechspace/fluxerly"const stop = new AbortController()process.once("SIGINT", () => stop.abort()) await using client = createClient({ token: process.env.FLUXER_BOT_TOKEN, })client.on("messageCreate", (message, signal) => { if (message.author.isBot) return if (message.content !== "!ping") return return client.messages.reply(message, "Pong!", { signal }) })const result = await client.run({ signal: stop.signal }) if (result.isErr() && result.error._tag !== "CancelledError") { console.error(result.error.message) }Effect is a TypeScript library for structured programs. The native entry point has the same features with typed failures and scoped cleanup, and it is never required
import { Config, Effect } from "effect"import { FluxerClient } from "@neontechspace/fluxerly/effect" const program = Effect.gen(function* () { const client = yield* FluxerClient yield* client.on("messageCreate", (message) => message.content === "!ping" ? client.messages.reply(message, "Pong!") : Effect.void, ) yield* client.run() })const client = FluxerClient.layerConfig({ token: Config.Redacted("FLUXER_BOT_TOKEN"), })export const main = program.pipe( Effect.scoped, Effect.provide(client), )JavaScript needs no TypeScript. TypeScript projects add TypeScript and @types/node. ES modules only, on the Node.js versions in the support guide
The open-source Fluxer bot that answers to you, from source code to server. It runs on Fluxerly.js through the SDK's Effect API, so every command below is this SDK at work
On the Website
An optional alternative to reaction roles, which work in chat. Each has its own switch and starts off, so a server can use either or both
general
Draws pixel towns and runs the Friday game night
showcase
A little painted toolbox, made over one long weekend
general
staff
Spamming links
direct message
Examples of what it does
It is written with the SDK's Effect API, the same one as these two samples
import { Effect } from "effect"import type { Client } from "@neontechspace/fluxerly/effect" export function readGuildOverview(client: Client, guildId: string) { return Effect.all({ guild: client.guilds.fetch(guildId), channels: client.channels.fetchAll(guildId), }, { concurrency: 2 })}import { Effect } from "effect"import type { Client } from "@neontechspace/fluxerly/effect" export function sendNotice(client: Client, channelId: string) { return client.messages.send(channelId, "The event starts in ten minutes").pipe( Effect.match({ onSuccess: message => ({ status: "confirmed" as const, messageId: message.id }), onFailure: error => ({ status: "failed" as const, error }), }), )}Each worker lives in an Effect scope, so nothing is left running when the bot stops
A job that fails is logged and the worker waits for the next one, so one bad job never takes the bot down
Effect keeps failures in the types, so each one gets its own plain reply instead of a crash
Answers from its backend are decoded with Effect Schema before the bot uses them
Profiles, showcases, levels, events and birthdays
A role picker or reaction roles, verification, newcomer roles and the member list
Cases, automatic rules, raid protection, logs and alerts
Announcements, greetings, custom replies and message cleanup
Tickets, the help desk and suggestions
Temporary voice rooms and finding a group
Check NeonFlux, apply presets, see activity and back up the server
Ping, help, away status, prefix and reply style
Forms update as the data changes, a change made elsewhere never overwrites your draft, and the audit log records every setting change
NeonFlux is a real Fluxer bot using Fluxerly.js, and public joining is on hold until it is ready. The source is open today
Each part comes with the checks and reviews behind it, and the test runs are public on GitHub
If a bot token can do it on Fluxer, you can do it today. Most routes have a named method, and a raw request on the same client reaches the rest, with the same rate limit handling and readable failures. The endpoint map lists every route and how to reach it
A dependable SDK can move when it needs to. Twice a day, if the SDK changed on the main branch and passed the check, the newest code goes to npm as a nightly build, so you can use a fix without waiting for a release. Another daily check compares Fluxer's API documents with the ones the SDK was built against
npm install --save-exact @neontechspace/fluxerly@nightly Frontier AI models review the code for security flaws, on top of the code scanning that runs on every push. A finding counts only with proof, such as a failing test or a traced path, and the maintainer decides every fix
Short answers to common questions, each with a link that goes deeper
More Answers in the FAQWhere Do I…?
Fluxer does not let bots register slash commands yet. Fluxerly will add them as soon as Fluxer supports them. Until then, Fluxerly commands are prefix commands, messages such as !ping that start with a chosen prefix, with typed arguments, guards, cooldowns, groups and generated help
No. The default entry point uses async and await with Results. The SDK uses Effect inside, and npm, pnpm and Bun install it automatically, while the Effect entry point is there for applications that already use it
No. Nothing needs to be switched on to receive events. A messageCreate handler gets messages as soon as the bot can read the channel
Node.js with ES modules only. JavaScript needs no compiler, and TypeScript projects add TypeScript with @types/node. The supported Node.js and TypeScript versions are listed in the support guide
Not yet. Voice connections and audio are deferred until after Fluxer's voice update. Voice state updates and member controls such as move, mute, deafen and disconnect work today
Voice ControlsThe SDK handles them. When Fluxer answers with HTTP 429, it waits the required delay and sends again within the request's deadline. Waits of a second or more are logged as warnings with the code ratelimit.wait
Yes. Releases follow Epoch Semantic Versioning, so only a new major version or epoch can break existing code. Minor and patch releases are safe to take
Support and CompatibilityExpected failures such as a missing permission, a network error or a rate limit stay visible in the return type, so they are handled where they happen. Programming mistakes still throw at once, and orThrow is there for code that prefers exceptions
The testing entry point creates a real client connected to an in-memory Fluxer. Tests deliver events, answer requests and check what the bot sent, with no network and no account
Testing GuideMessages are plain, read-only data, so no network call hides behind a property. A handler receives a reply function, and client.messages.send sends to any channel
Yes. AI helps write the code, and the maintainer decides what goes in. Code written with AI goes through the same check as every other change, and nothing is published unless that check passed
Quality, Speed and SecurityIt can, when the person using AI knows what to check. Tests and scanning look at the code itself, whoever typed it. Here the full check runs the whole test suite, and code scanning looks for security flaws
What the Full Check CoversYes. Nothing in the SDK needs an AI tool, and every bot can be written by hand. The SDK itself is built with AI, so it suits you best if you see the value in that, or are willing to give it a fair look
Quick StartNot when the person using AI knows what good code looks like. With that knowledge, AI-assisted code is more secure than code written by hand, because the time AI saves goes into more tests
How Security Is CheckedEditor documentation, test helpers and agent rules all come in the package you install
The documentation lives in the type declarations, so autocompletion in the editor shows the same text as the API reference
The /testing and /effect/testing entry points come with the SDK. A project from fluxerly init starts with its first test
fluxerly agents adds the SDK's rules to the AGENTS.md of a project that has the SDK installed, and the docs publish llms.txt. The rules are plain text a person can read too. Nothing here needs an AI tool
npx --no fluxerly agentsChoose your package manager, then add the SDK to a project or start a new one
npm install @neontechspace/Adds the SDK to an existing project
npx @neontechspace/Or start from a generated project with its first test
