AI MCP

Kanera's MCP server lets AI tools connect to Kanera through the Model Context Protocol. This gives an AI assistant structured access to your Kanera workspaces, boards, cards, assigned work, notes, comments, labels, custom fields, and activity.

Use MCP when you want an AI tool to help inspect, summarize, triage, or update real Kanera work instead of working from pasted screenshots or copied card text.

What MCP is

MCP is a standard way for AI clients to talk to external systems.

In Kanera, the MCP server sits between your AI client and the Kanera public API. The AI client asks the MCP server for Kanera data or asks it to perform an action. The MCP server then uses a workspace API key to make the allowed request to Kanera.

That means MCP access follows the same workspace model as Kanera's public API:

• API keys are created from Workspace settings -> API.
• API keys are scoped to the workspace where they were created.
• The API key scope controls what the AI can do.
• Private board access follows the API key's workspace access.
• Lists, labels, and custom fields are workspace-scoped, so AI tools can reason across boards in the same workspace.

Why use AI with Kanera

AI is most useful in Kanera when it has enough context to see work across boards.

Good uses include:

• Summarize the status of a board.
• Prepare a standup update from assigned work.
• Triage a person's work by urgency, stale cards, missing due dates, or missing metadata.
• Search cards, notes, comments, and attachment filenames.
• Draft a card from a Kanera note.
• Create cards from meeting notes or customer requests.
• Add comments that summarize an external discussion.
• Move cards, update due dates, assign people, or set labels when you explicitly ask the assistant to make changes.

Assigned Work is especially useful through MCP. Because Kanera can group assigned cards across boards in a workspace, an AI assistant can help a user or manager review workload without hopping between boards.

Create an API key

1. Open the workspace.
2. Go to Workspace settings -> API.
3. Create a workspace API key.
4. Choose the smallest scope that fits the job.
5. Copy the secret when it is shown.

The secret is shown once. Store it in the AI client or server environment you use for MCP.

API key scopes

Scope	Use it when
Read	The assistant should inspect, search, summarize, or report on Kanera work without changing anything.
Write	The assistant should be allowed to create cards, update cards, move cards, assign users, set labels, set custom field values, add comments, and edit team notes.
Admin	The assistant needs admin-level public API operations where supported. Use this sparingly.

Start with Read for exploration and status reporting. Move to Write only when the assistant should be able to change work in Kanera.

Connect with stdio

Use stdio when the AI client runs the MCP server locally.

Set:

Environment variable	Purpose
KANERA_API_KEY	The workspace API key copied from Kanera.
KANERA_PUBLIC_API_URL	The Kanera public API URL. Defaults to http://localhost:3001.

Example MCP client configuration:

{
  "mcpServers": {
    "kanera": {
      "command": "pnpm",
      "args": ["--filter", "@kanera/mcp", "stdio"],
      "env": {
        "KANERA_API_KEY": "kanera_live_...",
        "KANERA_PUBLIC_API_URL": "https://api.example.com"
      }
    }
  }
}

For a self-hosted Kanera deployment, use the public API URL that your machine can reach.

Connect with HTTP

Use HTTP when your Kanera deployment exposes the MCP service for remote AI clients.

The MCP HTTP endpoint is:

https://mcp.example.com/mcp

The client must send the workspace API key as a bearer token:

Authorization: Bearer kanera_live_...

The MCP service also exposes a health check:

https://mcp.example.com/health

Self-hosted deployments can expose the MCP service on its own hostname. Keep it behind HTTPS and only give API keys to AI clients you trust.

What the AI can read

Kanera exposes structured MCP resources for:

Resource	URI pattern	What it contains
Workspace	kanera://workspace/{workspaceId}	Workspace details and workspace-scoped setup.
Board	kanera://board/{boardId}	Board data, lists, visible cards, members, labels, and custom fields.
Card	kanera://card/{cardId}	Card detail, including labels, assignees, checklists, attachments, and linked notes.
Note	kanera://note/{noteId}	A Kanera note.

These resources let AI tools inspect Kanera data in a predictable JSON format.

Available actions

The MCP server provides tools for common Kanera workflows.

Area	Tools include
Discovery	List workspaces, open a workspace, list boards, open a board.
Search	Search cards, notes, comments, and attachment filenames.
Cards	Read card detail, create cards, update title/description/due date, move or reorder cards, archive or unarchive cards.
Assignments	Replace card assignees and list assigned work across a workspace or for one user.
Labels and fields	Replace card labels and set workspace custom field values.
Comments	Add a comment to a card.
Activity	List recent board activity and comments.
Notes	List notes, read a note, create notes, and update notes.

Write actions require a Write or Admin API key. If the key only has Read access, Kanera returns a permission error instead of making the change.

Built-in prompts

Kanera's MCP server includes prompts that AI clients can surface directly:

Prompt	Use
summarize_board_status	Summarize board progress, blockers, stale cards, and next actions.
prepare_standup_update	Prepare a concise standup update from assigned work and recent activity.
triage_assigned_work	Group assigned work by urgency and flag stale or underspecified cards.
draft_card_from_notes	Draft a card title and Markdown description from a Kanera note.

These prompts are starting points. You can also ask the assistant normal questions once the MCP server is connected.

Example requests

Try requests like:

• "Summarize this board and call out blockers."
• "Show me assigned work for this workspace grouped by urgency."
• "Find stale assigned cards with no due date."
• "Prepare a standup update for this user."
• "Draft a card from this note, but do not create it yet."
• "Create a card in the Intake list from these meeting notes."
• "Move the top priority bug to In Progress and assign it to Priya."
• "Add a comment summarizing the customer call."

For changes, be explicit about what the assistant should update. A good pattern is to ask for a draft or proposed plan first, then ask the assistant to apply it.

Security recommendations

Treat an MCP API key like any other integration secret.

• Use a dedicated API key for each AI client or MCP server.
• Use Read keys for reporting and exploration.
• Use Write keys only when the assistant should change Kanera data.
• Avoid Admin keys unless the workflow truly needs them.
• Rotate or delete keys that are no longer used.
• Do not paste API keys into chats, tickets, cards, notes, or shared docs.
• For shared AI clients, prefer a read-only key unless the team has agreed the assistant can take action.

Kanera records activity from API key actions, so changes made through MCP can be traced back to the API key name.

Troubleshooting

Problem	What to check
The AI client cannot connect.	Confirm the MCP command or URL is correct. For HTTP, use /mcp. For stdio, confirm the server can start locally.
The server says the API key is missing.	Set KANERA_API_KEY for stdio, or send Authorization: Bearer kanera_... for HTTP.
Requests return 401.	The API key is missing, invalid, revoked, or copied incorrectly.
Requests return 403.	The API key is valid but does not have access to that resource or does not have the required scope.
The assistant can read but cannot edit.	Use a Write or Admin API key for workflows that change Kanera data.
The assistant cannot find a board or card.	Check that the API key belongs to the right workspace and that the board is accessible to that workspace key.
Self-hosted MCP cannot reach Kanera.	Confirm KANERA_PUBLIC_API_URL points to the public API service and is reachable from where the MCP server runs.