MCP Use Cases & Workflows

Everything on this page is a real prompt you can type into a connected AI assistant. Single tools answer single questions; the real power is in compositions — the assistant chains tools together to run the workflows an SEO team actually runs.

Last updated: 2026-07-18

[!IMPORTANT] How writes work. When you ask your assistant to generate a brief or draft, Floyi’s own generation pipeline does the work — the exact same process as clicking Generate in the app. The result lands in your workspace with correct sections, internal links, and map position, and the assistant reads it back to you.

“What brands do I have and where do they stand?”

Good for starting a planning session, briefing a new team member, or checking which brand needs attention.

“Where am I weak and what should I write next?”

Three gap types come back: missing (no published content), weak coverage (content exists but covers the topic thinly), and low ranking (published but ranking past position 20). Missing is your net-new queue; the other two are your refresh queue.

“Create a brief for this topic, then draft it.”

Brief generation mirrors the app’s four-step modal: competitors, personas, keywords + knowledge base, then links. The last three inputs are optional at the API level, but the assistant should walk through every step and only skip one when you explicitly choose to:

Paid generation checks the available balance before starting. The assistant should confirm the displayed cost with you first. A pending brief can be cancelled through MCP and its reserved credits refunded; draft cancellation is not currently exposed.

An assistant can take a brand from zero to Topical Authority — creating the brand, generating personas, building the site architecture, running topical research, clustering keywords, and building the map:

1. "Create a brand for acme.com — analyze the site"        (brand foundation)
2. "Generate 3 buyer personas"                             (audience insights)
3. "Set the site type to SaaS. Build a SaaS Site + Resources, then generate the architecture"
4. "Set the core topic and run topical research"           (4 generation steps)
5. "Find duplicate topics in the research tree and merge them"
6. Optional: "Run AI Search Gaps — what do the AI engines cover that my tree doesn't?"
   (persona-grounded, cost confirmed first; review the gaps, add the keepers)
7. "Cluster the keywords"                                  (confirms cost first)
8. "Review the clusters — merge anything that overlaps"
9. "Generate the topical map"
10. Decide whether the optional content info snapshots (planning metadata) are useful, then generate them if wanted; confirm the resource URL prefix and generate URL slugs
11. "Advance to Topical Authority"

Things the assistant knows (and will tell you):

• Costs are surfaced before spending. Pipeline clustering enforces a confirmation flag; research, AI Search Gaps, content info, briefs, and drafts expose their cost or estimate so the assistant can get your approval before starting.
• Some steps are one-way. Building the map locks clustering — the assistant surfaces irreversible steps before taking them.
• Research topic steps are free within your plan’s allocation; keyword generation is the stage’s main credit spend.
• Site architecture has a scope choice. A site can continue as “[Site] + Resources”, “Resources Only”, or “[Site] Only” — the labels match your site type (e.g. “SaaS Site + Resources”). The site-only route goes directly to Topical Authority and skips topical research.
• Content info is optional planning metadata. It is useful for prioritization and stakeholder handoffs, but briefs and drafts do not depend on it. Make the choice before Topical Authority, where content-info generation becomes locked.

“Restructure my topical map.”

These commands drive the Authority Organizer, so they apply once the brand reaches Topical Authority. Before that, map-stage editing is limited to moving, renaming, and deleting clusters.

“How strong is my topical authority, and who’s beating me?”

“Load the client’s style guide into Floyi.”

Paste a client’s style guide document into the conversation and have the assistant extract it into Floyi’s Content Guide in one session:

Guide entries feed every subsequent brief and draft for that brand.

“Publish this finished draft.”

1. "Find content gaps for [brand]"
2. "Generate a brief for the top gap topic"
3. "Check the brief status" (assistant polls until complete)
4. "Generate a draft from the completed brief"
5. "Show me the full draft"
6. "Publish it to my WordPress site as a draft"

1. "Run a weekly report for [brand]"
2. "Show the pipeline overview with credits"
3. "Find the top 5 content gaps"
4. For the top 3, one at a time: generate the brief, review it, then continue to the next

Refreshing a page ranking #15 usually beats a net-new article.

1. "Find gaps for [brand], but only weak coverage and low rankings"
2. "For each, confirm the ranking URL and current position"
3. "Show internal linking opportunities for the top 3"
4. After updating: "Refresh authority and check for movement"

Before writing something new, check whether the map already targets it.

1. "Search my map for topics about '[keyword]'"
2. If a published topic already targets it -> refresh that page instead (Recipe 3)
3. If two unpublished topics overlap -> "Combine them into one topic"

1. "Who leads on authority — me or my competitors?"
2. "Find contested gaps: topics with high competitor presence where I'm missing"
3. "Which topics cite my domain in AI Overviews? Now the same for competitor.com"
4. "Diff the two lists — where are they cited and I'm not?"
5. Brief the highest-priority topic, review it, then continue through the remaining topics one at a time

1. "Where am I cited in AI Overviews? Where am I only mentioned?"
2. "Repeat for AI Mode, Gemini, and ChatGPT"
3. Mentioned-but-not-cited topics = citability gap — add stats, definitions,
   and quotable passages to those pages
4. Weekly: "Refresh authority and re-pull the citation counts"

Floyi generates deliberately one piece at a time — every brief and draft gets its own cost confirmation and a review before the next starts, so a sprint is a loop, not a batch:

1. "Find all missing topics in the '[cluster]' cluster"
2. "Check my credit balance and estimate the briefs for this list"
3. For each topic: "Generate the brief" -> review it -> curate weak sections
4. "Generate the draft from that brief" -> check its quality scorecard
5. "Stage it in WordPress as a draft" (or commit it to the blog repo)
6. Next topic

Agencies are natural MCP power users when their client brands live in the same Floyi workspace — one conversation can cover every brand visible to that connection.

1. "List all brands with their authority scores"
2. Per brand: "Run the weekly report"
3. "Show competitor movement for each"
4. "Pull AI citation wins to showcase"
5. "Assemble a client-facing summary email for each brand"

This can also become a scheduled digest when the MCP client supplies the scheduler and an email or Slack delivery integration. Floyi MCP provides the brand data and analysis tools; it does not send email or Slack messages by itself.

1. "Show me draft [title]"
2. "Get internal link suggestions for its topic"
3. Add the links in the editor (or verify the draft already includes them)
4. "Publish it as a WordPress draft"

The MCP server is built for strategy, production, and monitoring. A few things stay in the app on purpose:

For supported capabilities, connect and ask. The assistant should point you back to the app when a workflow requires visual review, account administration, credentials, or an unexposed import or analytics surface.