Documentation/How it works/Dashboard & MCP

Two surfaces, one project

Everything in a Rampify project (keywords, specs, crawl data, research) is one dataset with two front doors. The dashboard is where you see and curate it. MCP is how your AI does something with it, in the middle of a conversation, without you switching tabs.

The model#

Neither surface is a copy or an export of the other. Both read and write the same records, so a cluster your AI creates over MCP is on the Keywords page before you finish the sentence, and a spec you edit in the dashboard is what your coding tool sees on its next call.

DashboardMCP tools
PostureReview: you look at the dataAction: your AI works with the data
Whererampify.dev, a tab you visitClaude, Cursor, VS Code: wherever you already are
Best atOrientation, curation, spotting what's offResearch, generation, bulk changes, follow-through
ExampleScan clusters and notice comparison terms are weak“Add a cluster comparing us to our top 3 competitors”
Note

The mapping isn't 1:1 everywhere. Agent-invoked tools (the discovery_submit_* family) have no dashboard surface at all, and Project Settings deliberately has no MCP surface; credentials stay out of conversations.

Where the surfaces meet#

Each dashboard page pairs with a family of MCP tools over the same data:

Dashboard pageShared dataTool family
SpecsFeature specs, tasks, criteria, commitsFeature Specs
DiscoveryResearch sessions and findingsDiscovery Research Sessions
WebsiteURLs, crawl checks, issuesSite Analysis & Monitoring
KeywordsClusters and tracked keywordsKeyword Research & Management
Google SearchGSC performance and indexingSite Analysis & Monitoring
Business ProfilePositioning, audience, competitorsBusiness Profile & Competitors
ProductsProducts and their positioningBusiness Profile & Competitors

The bridge between them#

You don't have to remember the mapping; the product carries it. Every dashboard page's toolbar has an MCP handoff: a paste-ready prompt, pre-filled with the project context of the page you're looking at, that drops your AI into the same data over MCP. It's the one-click path from reviewing something in the dashboard to acting on it in a conversation. (Not connected yet? The same pill installs the connection; see Connecting.)

How these docs reflect it#

Every dashboard page ends with a From your AI tool section mapping what you just read to the tools that do it. In the other direction, tool pages carry a connected surface note pointing back to the dashboard page that views the same data, wherever the mapping is clean. Tools without one (agent-invoked bookkeeping, workflow utilities) genuinely have no dashboard counterpart, and that absence is informative too.