| name | mooboard-create-research-board |
| description | Creates a new Mooboard / Moobird qualitative research synthesis board on an infinite canvas. Use for interview insights, affinity mapping, evidence clusters, and finding→implication boards. |
| disable-model-invocation | true |
Mooboard Create Research Board
Create a new board under the Mooboard library as a research synthesis
board on an infinite board.
Core idea
Group evidence into insight clusters. Each insight has supporting quotes,
snippets, links, or images nearby. A “so what / implications” zone sits apart.
Do not rely on a layout script. Use multi-agent review.
Best use cases
- User interview synthesis
- Competitive teardown notes
- Literature / article synthesis
- Postmortem evidence boards
- Founder discovery research
Style rules
Usually includes:
- research question as title
- 3–7 insight clusters
- evidence notes/links/images under each insight
- opposing evidence / tensions called out
- implications + next experiments todo
It should not look like a pretty collage with no claims.
Layout character
Desired feel
- Analytical, scannable, argument-friendly
- Insights visually separated
- Evidence clearly subordinate to the claim
Default pattern
Affinity clusters
- Question header
- Insight islands across the canvas
- Implications column on the right
- Open questions / next research todo below implications
Spacing expectations
- Gap between insight clusters: 180–320px
- Gap within a cluster: 60–120px
- Hard minimum gap: 60px
Required multi-agent workflow
Agent 1 — Insight Architect
Drafts insight statements from the material.
Agent 2 — Evidence Mapper
Assigns quotes/links/images to insights.
Agent 3 — Layout Planner
Places clusters and implications.
Agent 4 — Collision Auditor
Checks spacing.
Agent 5 — Rigor Critic
Flags insights with weak/missing evidence.
Final Manifest Reviewer
Checks the written .mooboard.json:
- coordinates match the approved plan
- every image is
presentation: "interactive"
- every link is
previewMode: "card" with previewImage
- folder media matches referenced paths exactly
Workflow
Research Board Progress:
- [ ] 1. Confirm research question and board name
- [ ] 2. Gather notes, quotes, links, screenshots
- [ ] 3. Form insight clusters
- [ ] 4. Place evidence + implications + todo
- [ ] 5. Multi-agent rigor review
- [ ] 6. Write .mooboard.json
- [ ] 7. Remove orphan media; final review
Images and media (hard rules)
For every image on the canvas:
- use
type: "image"
- store media at the board root
- include
naturalWidth and naturalHeight
- preserve aspect ratio
- MUST set
presentation: "interactive" (preview tiles)
Never use presentation: "file" for images meant to be seen as previews.
Link cards must use:
previewMode: "card"- a real
previewImage URL
Audio / PDF / other attachments use type: "file" with presentation: "interactive" when the user should open/play them.
No orphan media. Every media file in the board folder must be referenced by a canvas object. Extra loose files get auto-imported as file cards and can destroy the layout.
Bounding boxes
Maintain a placement table of (id, type, x, y, width, height, region).
Treat images, links, todos, native text, geo, and file cards as blocking boxes.
Arrows may cross freely and are not blockers.
Never allow accidental overlap between blocking boxes.
Anti-patterns
- Insights with no evidence nearby
- Everything in one blob
- Implications that ignore the evidence
presentation: "file" images or orphan media
Camera and envelope
{
"version": 1,
"createdAt": 0,
"lastOpenedAt": 0,
"objects": [],
"camera": { "x": 400, "y": 300, "zoom": 0.4 },
"background": "default",
"showGrid": true
}
Use epoch-ms timestamps. Object IDs: UUIDs for moobard objects; shape:<id> for native shapes.
Output to the user
Report:
- Absolute board path
- Layout pattern used and key region counts
- Counts of images, links, notes/todos, and other objects
- Any skipped URLs or assets
- That they should open or refresh the folder in Mooboard to see it
Do not claim the app rendered the board unless that was observed.
