MCP Server

What is MCP?

The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external tools and data sources. Kansov's MCP server exposes your workspace data through 23 tools that AI clients can call on your behalf — reading your roadmap, creating insights, searching across modules, and more.

Think of it as giving your AI assistant direct access to Kansov, so instead of copy-pasting data between apps, you can just ask: "What are my high-priority todos?" or "Create an insight about the pricing feedback from today's call."

Getting Started

Kansov uses OAuth 2.0 to authenticate MCP connections. You'll create an OAuth client in Kansov, then configure your AI assistant to connect using that client.

Kansov's MCP server works with any MCP-compatible AI client including ChatGPT, Claude Desktop, Cursor, and others. Refer to your AI client's documentation for specific MCP setup instructions.

---

Available Tools

Kansov exposes 23 tools organized into three categories: Read, Create, and Update. All tools respect your OAuth scopes and workspace permissions.

Read Tools

These tools retrieve data from your workspace. They don't modify anything.

Tool	Description	Key Parameters
list_insights	List customer feedback, pain points, and observations	limit, search, status, createdAfter, createdBefore
list_ideas	List ideas with RICE scores and linked goals	limit, search, status, createdAfter, createdBefore
list_todos	List personal tasks and action items	limit, search, completed, priority, createdAfter, createdBefore
list_features	List features with progress and linked goals	limit, search, status, createdAfter, createdBefore
list_knowledge	List knowledge base documents, notes, and research	limit, search, createdAfter, createdBefore
list_goals	List company and product goals with vision alignment and linked idea/feature counts	limit, search, level, status, visionId
list_roadmap	List the product roadmap — ideas organized by Now / Next / Later buckets	limit, timeframe, productArea, search
list_meetings	List meetings with agendas, participants, and recurrence	limit, search, filter (upcoming / past / all)
search	Search across multiple modules simultaneously	query, modules, limit

Create Tools

These tools create new items in your workspace. AI clients will ask for your confirmation before executing.

Tool	Description	Key Parameters
create_insight	Capture customer feedback or observations	title (required), description, source, customerName, productArea
create_idea	Capture a potential solution or feature concept	title (required), description, productArea
create_todo	Create a task or action item	title (required), description, priority (low/medium/high/urgent), dueDate
complete_todo	Mark a todo as completed or incomplete	id (required), completed (true/false)
create_feature	Add a new feature to the roadmap	title (required), description, status (planning/in-progress/done)
create_knowledge	Add a document or note to the knowledge base	title (required), content (required)
create_meeting	Schedule a new meeting	title (required), agenda, meetingDate, recurrence

Update Tools

These tools modify existing items. All require the item's id — use a read tool first to find it.

Tool	Description	Updatable Fields
update_insight	Update an existing insight	title, description, status, source, customerName, productArea
update_idea	Update an existing idea	title, description, status, productArea, impact, effort, reach, confidence
update_todo	Update an existing todo	title, description, priority, dueDate
update_feature	Update an existing feature	title, description, status, productArea
update_knowledge	Update a knowledge base item	title, content
update_meeting	Update a meeting's title, agenda, or date	title, agenda, meetingDate

---

Sample Responses

Here's what the AI assistant sees when it calls Kansov tools. These responses include strategic context like linked goals and vision alignment.

list_ideas

Ideas include RICE scores and the full goal hierarchy chain — so the AI understands how each idea connects to your strategy.

{ "items": [{ "id": "idea-abc123", "title": "Add dark mode support", "status": "shortlisted", "productArea": "UX", "reach": 8, "impact": 6, "confidence": 9, "effort": 4, "linkedGoals": [{ "id": "goal-xyz", "title": "Improve user retention by 20%", "level": "company", "status": "in_progress", "parentGoal": null, "vision": { "id": "vis-001", "title": "Best-in-class user experience" } }] }], "count": 1 }

list_goals

Goals show the full hierarchy plus how many ideas and features are linked to each goal.

{ "items": [{ "id": "goal-xyz", "title": "Improve user retention by 20%", "level": "company", "status": "in_progress", "metric": "Monthly retention rate", "target": "80%", "timeframe": "Q2 2026", "parentGoal": null, "vision": { "id": "vis-001", "title": "Best-in-class user experience" }, "linkedIdeaCount": 5, "linkedFeatureCount": 3 }], "count": 1 }

search

Cross-module search returns results from multiple modules with a short preview.

{ "query": "onboarding", "results": [{ "type": "insight", "id": "ins-101", "title": "Users drop off during onboarding step 3", "preview": "In the last 2 weeks, 40% of new users abandon..." }, { "type": "idea", "id": "idea-202", "title": "Simplify onboarding to 3 steps", "preview": "Reduce the current 7-step onboarding..." }], "count": 2 }

---

Example Prompts

Here are real prompts you can try with your connected AI assistant. Each one maps to specific Kansov tools.

What You Ask	Tools Used	How It Works
"List my top 5 insights from this week"	list_insights	Calls with limit: 5 and createdAfter set to this week
"Create a todo to review the Q2 roadmap by Friday"	create_todo	Calls with title, dueDate, and priority: "medium"
"What's on our roadmap for the Now bucket?"	list_roadmap	Calls with timeframe: "now"
"Show me all company-level goals"	list_goals	Calls with level: "company"
"Search for anything related to onboarding"	search	Calls with query: "onboarding" across all modules
"Show me upcoming meetings and their agendas"	list_meetings	Calls with filter: "upcoming"
"Mark my todo 'Fix login bug' as done"	list_todos → complete_todo	Searches for the todo, then marks it complete
"Change idea 'Dark mode' status to in-progress"	list_ideas → update_idea	Finds the idea, then updates its status
"List all high-priority todos that aren't done"	list_todos	Calls with priority: "high" and completed: false
"What features are currently in progress?"	list_features	Calls with status: "in-progress"
"Create an insight about pricing feedback"	create_insight	Calls with title, source: "customer call"
"Add a topic to the Weekly Product Sync"	list_meetings → update_meeting	Finds the meeting, appends to its agenda

---

OAuth Scopes

When creating an OAuth client, you choose which scopes to grant. Each scope controls access to specific tools.

Scope	Tools Granted
insights:read	list_insights
insights:write	create_insight, update_insight
ideas:read	list_ideas
ideas:write	create_idea, update_idea
todos:read	list_todos
todos:write	create_todo, complete_todo, update_todo
features:read	list_features
features:write	create_feature, update_feature
research:read	list_knowledge
research:write	create_knowledge, update_knowledge
meetings:read	list_meetings
meetings:write	create_meeting, update_meeting
goals:read	list_goals
roadmap:read	list_roadmap

Endpoints

These are the key endpoints for connecting to Kansov's MCP server. Replace your-domain.com with your actual Kansov URL.

Endpoint	Purpose
/mcp	MCP transport endpoint — where AI clients send tool calls
/api/oauth/authorize	OAuth authorization endpoint — user consent screen
/api/oauth/token	OAuth token endpoint — exchange auth code for access token
/.well-known/oauth-authorization-server	OAuth 2.0 server metadata (RFC 8414) — auto-discovery of all OAuth URLs
/mcp/.well-known/oauth-protected-resource	MCP protected resource metadata — auto-discovery for MCP clients
/.well-known/mcp.json	MCP server discovery manifest — transport type, endpoint URL, auth info

Security & Permissions

• OAuth 2.0 with PKCE — All connections use the authorization code flow with PKCE for security.
• Scoped access — Each OAuth client only has access to the scopes you grant. You can revoke access at any time.
• Workspace-level isolation — Tools only return data from your workspace. You can never access another workspace's data.
• Role-based access — MCP tools respect your workspace role (Admin, PM, Contributor, Viewer). A Viewer connecting via MCP still only has read access.
• Write confirmation — AI clients are instructed to confirm before executing write operations (create, update, complete).