Content briefs

A brief is the planning document that tells the AI what to write. Every piece of content starts with a brief.

Why briefs matter

Without a brief, AI-generated content is generic. With a well-formed brief, the output is specific, on-brand, and targeted to the right audience and platform.

A brief answers: What is this about? Who is it for? What platform? What angle? What should the reader do next?

Brief fields

Field	Required	Description
topic	Yes	The main subject of the content
platform	Yes	Where this will be published
audience	No	Which audience segment to target
angle	No	The specific take or frame for the topic
tone	No	Tone override (default: from engine config)
targetLength	No	Approximate word or character count
keyPoints	No	Bullet list of points to include
callToAction	No	What to ask the reader to do
references	No	Links, quotes, or data to incorporate
context	No	Any additional notes for the AI

Creating a brief in the UI

Most briefs are created for you: every piece on a project canvas carries one, written when the project assembles. To write one directly, open the Generate page (it lives at /generate; projects are the sidebar's front door):

1. Write the prompt: the first line becomes the topic, and the whole prompt (plus any attached reference files) becomes the brief
2. Pick the channels to generate for
3. Click Generate

Reactor creates the brief and runs it immediately, streaming one draft per channel. (The New menu on the Content page is different: it creates blank drafts or imports documents, with no brief involved.) Briefs created over MCP or by a manual-mode project enter a pending state instead and generate when you run them.

Creating briefs via MCP

{
  "tool": "create_brief",
  "arguments": {
    "topic": "Why context windows don't solve memory in AI agents",
    "platform": "linkedin",
    "audience": "Developer advocates",
    "angle": "Counterintuitive take: bigger context != better memory",
    "keyPoints": [
      "Context windows are volatile, not persistent",
      "Real memory requires retrieval, not recency",
      "Agents that chunk into context fail at scale"
    ],
    "callToAction": "Ask what retrieval strategy they're using"
  }
}

The create_brief tool's full schema is advertised by the MCP server and read directly by your client. See the MCP overview for how this works.

Generating from analytics

The suggest_brief_from_analytics MCP tool analyzes your top-performing content and suggests a new brief that follows the same patterns. This is useful for scaling what's already working.

Brief status

Status	Meaning
pending	Brief created, ready to generate
generating	AI is currently writing content from this brief
done	Content has been generated
failed	Generation failed; brief can be retried