Figma and Linkeddit in Claude: FigJam competitive analysis boards and battlecards

The Figma connector is Figma's official MCP server at https://mcp.figma.com/mcp, and its listing shows 9 tools, 6 reads and 2 writes. It is built mainly for design to code. Its help centre says it gives agents the design context to generate code from Figma files, and lets agents write native Figma content back to the canvas. The listing names generate_diagram as the tool that creates FigJam diagrams.

It knows nothing about what your competitors shipped this week or what ChatGPT says about your category. Linkeddit is the source of facts. Compete tracks competitor moves and stores their customers' reviews. Answer Radar asks the five engines your buying questions and records who is named and recommended. Both come with one plan at $60 a month.

Claude reads the data, decides what belongs on the board, and asks the Figma connector to draw it. Linkeddit writes no copy or design of its own.

Figma's help centre describes the generate_diagram tool as making a FigJam diagram from a description or from Mermaid syntax, and lists flowcharts, Gantt charts, state diagrams, sequence diagrams, architecture diagrams and entity relationship diagrams. It adds that you can make sure the agent uses it by including using the Figma MCP generate_diagram tool in the prompt.

The same help page says the remote server can write native content to Figma Design, FigJam and Figma Slides, needs a Full seat to write outside your drafts, and is free during a beta period that will eventually become usage-based and paid. Check Figma's current terms before building a routine on it.

A matrix or table is not in the diagram list, so the classic competitor grid still comes from a template.

In practice that means three useful boards. A Gantt chart of competitor launches by week, a flowchart of a buyer's path from a pricing change to a switch, and an architecture diagram of which competitor covers which part of your stack. Each one is drawn from signals Claude has already read, so every node can carry the source link behind it.

The first is a FigJam competitive analysis: Claude reads the weekly Compete brief and draws a timeline or a buyer flow with a source link behind each node. Open the guide for the exact prompt and a real run.

Both guides use the same pattern. Claude reads first, shows you a plan with the source behind each node, and only then draws or proposes text. Neither guide lets Linkeddit write to Figma, and neither claims anything the Figma connector's own documentation does not say it can do.

The second is a FigJam battlecard, refreshed from competitor signals and stored reviews. Claude reads the card through the Figma connector, proposes text for the data boxes, and reports which review sites failed to load.

A board of competitor names is not the goal. A Claude user on Reddit put it this way: "Real competitive intelligence is not a list of names." The workflows here put data behind the board.

Answer Radar's results tool returns visibility metrics per segment: presence, recommendation, share of voice, position, sentiment and citation share, plus the domains the engines cited. Claude can read those and turn them into a flowchart or an architecture diagram, such as which engines name a competitor first and which sources they cite for it.

Figma's tool does not list a chart type, so do not expect a bar chart of share of voice from generate_diagram. If you need numbers in a grid, ask Claude to write the table in its reply and paste it into a FigJam table, or use a template.

Linkeddit measures through provider APIs with grounded web search, and says in each result that this is a controlled proxy, not an exact copy of what a person sees in consumer ChatGPT or Google. Put that line on the board.

Read every node against the source. Ask Claude to read the finished board back with get_figjam and compare it with the Linkeddit data, since a drawing step can shorten or drop a line.

Check what is missing as well as what is there. Linkeddit's weekly brief for its own test project showed 5 competitors monitored, and one of them, plausible.io, was still pending with a count of 0 that week. Linkeddit reports a status for each competitor and review site, so a failed fetch shows as a failure and not as a quiet competitor. Mark thin evidence on the board.