Skip to main content

Content · build guide

Content Brief

Writer-ready SEO briefs from a target keyword in under 20 minutes instead of three hours.

Path A · Manual extracts Path B · MCP integration Guardrail-first

Content Brief starts from a target keyword and ends with a complete, writer-ready SEO brief: search intent classification, audience and angle, a structured outline with a word-count target, internal link suggestions pulled from your own site, and a handoff prompt for Claude or Cursor. Every brief lands in the same standardized .md template, so writers stop guessing what a good brief looks like.

Briefing is the bottleneck in most content operations, which is why this compounds. A brief that took a strategist around three hours now takes under 20 minutes, so the same team covers more of the keyword map without lowering the bar. And because every brief follows the same template and logs its sources, the process is inspectable, repeatable, and survives staff turnover.

What it moves: Brief production time drops from roughly three hours to under 20 minutes per brief, so a keyword can go from queue to writer-ready the same day.
Path A · Manual

This version runs on keyword exports, a Google Sheet, and a saved prompt pack in Claude — no APIs, no middleware. It suits teams producing a handful of briefs a week who want the templates and structure proven before wiring anything together. You gather the data; the prompts do the assembly; an editor signs off.

Path B · Integrated

Here the classify → SERP → outline → brief loop runs as an agent workflow. Keyword data, Search Console, and your site inventory connect through read-scoped API or MCP access, and the agent works a queue instead of waiting on a strategist. It suits teams briefing at volume that already have a documented content process and a real editorial gate.

Prerequisites.

A keyword research tool you already use (Semrush, Ahrefs, or similar) with export or API access
A documented content process: who requests briefs, who writes, who approves
An inventory of your site's URLs (sitemap or a Screaming Frog crawl) for internal-link suggestions
Read access to Google Search Console for your domain, to sanity-check intent against real queries
New to the stack? Set up your environment first — data access, workspace, integration platforms, and the agent layer are covered once in Environments & tooling.

Build it with manual data extracts.

This version runs on keyword exports, a Google Sheet, and a saved prompt pack in Claude — no APIs, no middleware. It suits teams producing a handful of briefs a week who want the templates and structure proven before wiring anything together. You gather the data; the prompts do the assembly; an editor signs off.

  1. Set up the brief template and intent rubric

    Create a .md brief template with fixed sections: audience, angle, outline, word-count target, internal links, and a handoff prompt block. Write a one-page rubric for classifying keywords as informational, commercial, or navigational. Get your writers to approve the template before anything else — it is the contract the whole workflow fills in.

  2. Export your keyword queue

    Export target keywords from your keyword tool into a Google Sheet with volume and difficulty columns. Add a status column (queued, briefed, drafted, published) and an owner column. This sheet is now the single queue every brief comes from.

  3. Capture the SERP by hand

    For each queued keyword, run the search and record the top 10 results in a SERP analysis tab: titles, recurring H2 themes, approximate word counts, and whether a featured snippet is present. Ten minutes of structured looking beats an hour of unstructured browsing.

  4. Run the prompt pack in Claude

    Paste the keyword, your intent classification, and the SERP notes into a saved Claude prompt that outputs the brief in your exact template. Save the result as a .md file in a shared briefs folder. Keep the prompt wording in a dated doc so you can see which revision produced which brief.

  5. Add internal links from your URL inventory

    Check your sitemap sheet or Screaming Frog crawl export and pick two or three genuinely relevant internal links. Paste them into the brief's internal-links section with one line of anchor guidance each.

  6. Editor review and writer handoff

    An editor reads every brief before it reaches a writer — checking angle, outline, and links against the rubric. Log the date, keyword, prompt used, and approver in the tracking sheet, then flip the status to briefed and hand off with the Claude/Cursor prompt attached.

Cadence: Batch weekly against the keyword queue; expect roughly 20 minutes per brief once the template and prompt pack are settled.

Integrate it with MCP connections.

Here the classify → SERP → outline → brief loop runs as an agent workflow. Keyword data, Search Console, and your site inventory connect through read-scoped API or MCP access, and the agent works a queue instead of waiting on a strategist. It suits teams briefing at volume that already have a documented content process and a real editorial gate.

Your keyword research tool (read-only API access or an MCP server for it)
Google Search Console (read-only) for query and ranking data
Google Sheets or your project tracker — the brief queue and audit log
Google Drive or your docs store, where finished .md briefs land
Slack or email for pending-approval notifications
  1. Wire the data sources, read-only

    Connect your keyword tool and Google Search Console to the agent using read-only credentials, and point it at a sheet or export of your site's URL inventory. The agent gets no CMS login and no publishing access of any kind — it reads data and writes brief files, nothing else.

  2. Define the queue and the trigger

    The brief queue lives in a Google Sheet or your project tracker. A new row with status queued triggers a run — via n8n, Make, or Zapier, or on a scheduled agent check. Anyone on the team can request a brief by adding a keyword; nobody has to know how the machinery works.

  3. Let the agent run the brief loop

    For each queued keyword the agent classifies intent, pulls SERP and keyword data, checks Search Console for related queries you already rank for, generates the outline and word-count target, and selects internal links from your URL inventory. It writes the finished brief into your .md template in the shared briefs folder.

  4. Record every run

    Each run appends an audit entry: keyword, data sources pulled, prompt used, timestamp, and output file location. If a brief looks off, you can trace exactly what the agent saw and regenerate it against the same inputs.

  5. Hold the approval gate

    The agent posts each finished brief as pending — a Slack message or a status flip in the queue. An editor approves it or sends it back with notes; only approved briefs move to writer handoff with the Claude/Cursor prompt attached. The agent never advances a brief past the gate on its own.

Run it safely.

Google Sheets — the keyword queue, status tracking, and SERP worksheet in the manual path
Claude — runs the prompt pack by hand, or the full agent loop via MCP servers
Semrush or Ahrefs — keyword volume, difficulty, and SERP data by export or API
Screaming Frog — crawls your site to build the internal-link inventory
Google Search Console — real query data to validate intent classification
n8n, Make, or Zapier — triggers the agent when a keyword enters the queue
  • No CMS access, period: the agent reads keyword, SERP, and Search Console data and writes only brief files — no publishing scope of any kind
  • Every brief passes an editor approval gate before writer handoff — nothing enters the content calendar unreviewed
  • Per-run audit log records keyword, data sources, prompt used, and output file, so any brief can be traced and regenerated
  • Fully reversible: briefs are files in a pending folder — rejecting one discards a document, never published content

Want new build guides as they ship?

Optional — every guide on this site is free either way. Leave your details and we'll email you when new automations join the library, nothing else.

We only email when there's something worth building — every email has a one-click unsubscribe, and you can withdraw consent anytime. Consent requested by Campaign Automation AI, Caledon, Ontario · hello@campaignautomation.ai.

Prefer it built for you?

We stand this automation up as a governed sprint on the stack you already run — scoped, built, and handed over with the guardrails configured. No new platform, no migration.

Start with the Audit → Request proposals →