Nuelink Developer Documentation

Version: 1.6 API contract: 1.6.0 Base URL: https://app.nuelink.com/api/public/v1

Nuelink offers four ways to publish programmatically. They share the same account, the same brands, the same collections, and the same channels, the difference is who’s driving.

Four surfaces, one Nuelink

REST API

HTTP endpoints for developers building integrations, automations, or custom apps. Send a POST request, get back a post ID. Works from any language that can make HTTP calls.

→ Start with the API Quick Start

MCP Server

Connect Nuelink to Claude, ChatGPT, Cursor, Manus, Codex, or any other MCP-compatible AI assistant, and publish posts in natural language. OAuth is the recommended path for new connections, with no API key embedded in the server URL.

→ Start with the MCP Quick Start

CLI

nuelink-cli on npm. Publish from your terminal, a shell script, or a CI job. Every API mutation supports --dry-run, and --json makes output pipe cleanly into jq.

→ Start with the CLI Quick Start

Agent Skills

A skills pack for safe Nuelink CLI workflows: discover first, preview mutations, and default ambiguous publishing intent to DRAFT. The installer is tested with Claude Code and Codex; compatibility with other runtimes depends on their supported format and install paths.

→ Start with Agent Skills

You can use all four. Across the current public operation set, anything you can do through the REST API, MCP server, or CLI can be done through the other operational surfaces. Authentication, parameter naming, output formatting, and safety controls differ by surface.

What’s new in 1.6

Change	Where
Posts show their channels: every post you list or update returns the channels it publishes to	Endpoints
Automations return their settings: addPostsAsDraft and refreshRate	Endpoints
Automatic text splitting is on by default: long text becomes a thread unless you send autoThreadText: false	Threads
Auto-comment delays are in minutes, at least 1, instead of seconds	Global auto-comment
CLI 1.6.5 shows post channels and automation settings, refuses redirects, and takes --comment-delay in minutes	CLI Commands

1.5 added editing collections, captions, and automations. See the Changelog for every release.

Base URL change

The canonical base URL is https://app.nuelink.com/api/public/v1. The previous https://nuelink.com/api/public/v1 host remains supported for compatibility, but new and updated integrations should use app.nuelink.com. A future deprecation and removal timeline for the previous host will be announced separately. The CLI already defaults to the canonical URL.

How Nuelink is organized

Before you make your first call (or your first prompt), it helps to understand the three-level hierarchy Nuelink uses:

Account
  └── Brand (workspace, e.g. "Fernwood Botanicals")
        ├── Channels     (the connected social accounts: IG, X, TikTok, …)
        ├── Collections  (content folders grouped by theme or campaign)
        │     └── Posts  (social media posts)
        ├── Automations  (feeds that create posts for you, on a schedule)
        └── Media        (the brand's uploaded image and video library)

Brand

A fully-isolated workspace. Each brand has its own channels, collections, automations, media library, and insights, so you can manage multiple companies or projects without their data mixing. Most accounts have one brand; agencies and power users typically have several.

Collection

A content folder inside a brand, typically organized by theme or campaign (e.g. “Summer Campaign”, “Product Launches”, “Customer Stories”). Each collection has a set of default channels and weekly time slots (aka Queue). When you create a post inside a collection, it publishes to the collection’s default channels. A brand’s schedule is every collection’s time slots, read together.

Channel

A single connected social account, e.g. @fernwoodbotanicals on Instagram or your company LinkedIn page. Channels belong to a brand.

Automation

A feed that turns into posts. Point an automation at an RSS feed, a YouTube channel, a Shopify store, a Substack, or another supported source, and Nuelink imports new items into a collection as they appear.

Creating a post always follows the same path

Pick a brand → pick a collection inside it → POST the post to that collection.

Channels are listed as a separate endpoint so you can inspect what’s connected to the brand, but you don’t pass channel IDs when creating a post, the collection’s configuration determines which channels receive it.

Which surface should you use?

	REST API	MCP Server	CLI	Agent Skills
For	Developers writing code	Anyone using an AI assistant	Terminal and CI users	Coding agents
Interface	HTTP requests	Natural language	Shell commands	Agent instructions
Best for	Custom apps, integrations, scheduled jobs	Drafting, scheduling, day-to-day posting	Scripts, pipelines, one-off ops	Teaching an agent safe defaults
Auth	API key in Authorization header	OAuth; deprecated query-key compatibility for migration	API key via flag, env var, or saved config	Inherits the CLI’s auth
Workflow	Code → request → response	Prompt → assistant → tool calls	Command → response	Task → skill → CLI commands

If you’re building a piece of software that publishes posts (a content scheduler, a CMS integration, an automation pipeline) use the API.

If you want to publish posts from an AI assistant you’re already using to draft them (Claude, ChatGPT, Cursor, etc.) use the MCP Server.

If you want to publish from a shell script, a cron job, or GitHub Actions, use the CLI.

If you want your coding agent to run those CLI commands correctly and safely without you supervising every step, install the Agent Skills pack.

If you’re not sure, start with the API Quick Start — it needs only an API key. The MCP Quick Start is faster still if your assistant supports OAuth. You can layer the others on top later.

What the platform covers today

The REST API, MCP server, and CLI cover the same 24 operations. Agent Skills guide agents through those CLI workflows:

1. Get the current profile
2. Validate a token
3. List brands
4. Read a brand’s weekly schedule
5. List channels
6. List collections
7. Create a collection
8. Edit a collection
9. Add a channel to a collection
10. Remove a channel from a collection
11. Add a time slot to a collection
12. Delete a time slot
13. List automations
14. Create an automation
15. Pause or activate an automation
16. List media
17. Upload media
18. List posts in a collection
19. List posts across a brand
20. Create a post
21. Edit, reschedule, re-queue, or draft a post
22. Move posts to another collection
23. Delete a post
24. List published results, with engagement insights

Any post can be moved or deleted, and posts that haven’t been published yet can have their caption edited, or be rescheduled, re-queued, or drafted. A post’s media and other options can’t be edited through the API, and polls can’t be edited at all. Collections can be edited but not deleted, automations can be paused and activated but not otherwise changed, and media is create-and-read only. Webhooks and analytics beyond per-post engagement are not part of the current public operation set.

The safe publishing model

Every surface creates posts through the same four modes. The order below is also the order of risk:

Mode	Effect	When to use it
DRAFT	Saves without scheduling or publishing	First integrations, and any time intent is unclear
QUEUE	Adds the post to the collection’s next queue slot	Day-to-day publishing on a set schedule
SCHEDULE	Publishes at scheduledAt in the brand’s timezone	A specific, confirmed date and time
IMMEDIATE	Attempts to publish now	Only after explicit confirmation

Build against DRAFT first. Point your integration at a collection, create drafts, and check them in the dashboard. Move to QUEUE or SCHEDULE once you trust the output. Nothing reaches an audience until you choose a non-draft mode.

The same order applies after a post exists. Turning it back into a DRAFT is always the safe change. Deleting is a separate sensitive action, refused until the account owner turns it on.

Documentation map

API

• Quick Start - your first API draft in 5 steps.
• Authentication - bearer tokens and how to handle them.
• Rate Limits - 60 per minute per IP, 30 per user on resource endpoints, and how to back off.
• Errors - the four error shapes, and when a retry is safe.
• Pagination - page-number pagination across surfaces.
• Endpoints - full reference for all 24 endpoints.
• Platform-Specific Options - stories, threads, polls, comments, Instagram collabs, YouTube playlists.

MCP Server

• Overview - what the MCP server does and when to use it.
• Quick Start - connect Nuelink to your AI assistant in under 2 minutes.
• Authentication - OAuth and API keys.
• Available Tools - the 24 tools the server exposes.
• Example Prompts - copy-paste prompts for common workflows.

CLI

• Overview - what the CLI is for, and how it behaves.
• Quick Start - install, authenticate, first draft.
• Commands - every command and flag.

Agent Skills

• Overview - the skills pack and what it teaches your agent.

Resources

• Changelog - releases for the API, MCP, CLI, and Skills.