Mapping for agents
Your agent shipped exactly what the ticket said (and broke three things it didn't). CardBoard carries the context the ticket couldn't, so it builds what you meant.
Works with Claude, Claude Code, and anything that speaks MCP.
Point your agent at a plan and most tools hand it a grid of pixels and a wall of text. So it guesses, and you clean up after it.
CardBoard maps are typed: the method is in the map. Your agent reads the rules before it acts, and gets held to them when it writes.
What's live, and what's next.
- MCP connection
Claude and Claude Code, over MCP.
- Read the map
The archetype's rules and anti-patterns, plus cards, connections, comments, and tracker sync status.
- Propose card edits
Description and acceptance criteria, proposed as ghosts.
- Propose connections & comments
Edges and decision notes, proposed as ghosts.
- Start a board from Mermaid
Paste a Mermaid diagram into a new board; CardBoard picks the matching archetype and lays it out. (Human-driven — see Quickstart.)
- Agent-created maps & cards
Creating new surface over MCP. Today agents propose against maps a human placed.
- Outcomes as first-class
Cards already carry the laddering; reading and proposing directly against outcomes is next.
- More map types
Prioritization matrix, Wardley maps.
- More clients
ChatGPT, Gemini, any MCP-speaking agent.
Skip the blank-canvas cold start.
Ask Claude for a Mermaid diagram of your flow, story map, or tree. Paste it into a new board and CardBoard reads the diagram type, picks the matching map archetype, and lays it out structured and connected. A flowchart comes in as a flow, a tree as a tree, each already carrying its own rules. From there, point your agent at the map and it proposes against real structure — the same structure the import just set.
The map teaches your agent the method.
A story map isn't a grid of boxes. It's a grammar: activities on top, stories beneath, sequence left to right. CardBoard hands that grammar over up front — how the board is shaped, and the mistakes people make on it.
So your agent proposes work the way someone who knows the method would, not a card dump.
And holds your agent to it.
Reading the rules is easy. Following them is the work.
Drop a card between two lanes, hang a node with no parent — it won't take. The rules your agent read are the rules it's checked against, because both come from one definition. An archetype can't advertise one contract and enforce another.
Your agent reads the structure. It doesn't guess it.
No squinting at coordinates to guess what connects to what. The read is the real structure: archetype rules, outcomes, connections, comments, and which cards are synced to a tracker. Enough to reason about the work instead of pattern-matching the text.
Connect a canvas tool to an agent and it has to infer what your layout meant, and that inference frays as the board grows. CardBoard declares the structure instead, so there's nothing to infer.
That edge compounds as the work gets bigger.
Agents get a voice, not a vote.
Agents draft cards and suggest where they go on the map. Nothing's real until you accept it. You stay the author; the agent just makes the next good idea cheap.
Every move is on the record, stamped with the person who prompted it, and only workspace members can put an agent to work — not anyone holding a share link.
The agent proposes on the map you built. It can't spin up its own — creating new surface isn't even in its hands.
Found in 3 support threads and a Jira epic. Sits under Check out.
ghost · awaiting a human's accept
What you'd actually ask it.
“Draw this onboarding flow as a Mermaid diagram.”
Claude returns a flowchart. You paste it into a new board; CardBoard reads the diagram type, picks the flow archetype, and lays it out connected. Then you point your agent at the map and it works against real structure.
“Add descriptions and acceptance criteria to the cards on the sitemap.”
The agent reads the map and proposes description + acceptance-criteria edits on existing cards as ghosts. You accept the ones that fit, edit any you want, and the agent re-reads your changes.
“Read the checkout story map and plan the guest-checkout feature.”
The agent reads the map's structure — backbone, stories, connections — and proposes new connections and card edits that fit the grammar, each as a ghost for you to accept.
Point your agent at a map.
A board that already knows the rules.
Works with Claude, Claude Code, and anything that speaks MCP.