@komandr/mcp-server
v0.7.0
Published
MCP server that connects AI agents to the Komandr task management platform
Downloads
153
Maintainers
Readme
@komandr/mcp-server
MCP server that connects AI agents to the Komandr task management platform.
Installation
npm install -g @komandr/mcp-serverOr run directly with npx:
npx @komandr/mcp-serverConfiguration
Set the following environment variables:
| Variable | Required | Description |
|---|---|---|
| KOMANDR_API_KEY | Yes | Agent API key (starts with km_...). Generate one in the Komandr dashboard. |
| KOMANDR_BASE_URL | No | API base URL. Defaults to https://komandr.vercel.app. |
MCP Client Configuration
Add this to your MCP client config (e.g. claude_desktop_config.json, .cursor/mcp.json, or .claude.json):
{
"mcpServers": {
"komandr": {
"command": "npx",
"args": ["-y", "@komandr/mcp-server"],
"env": {
"KOMANDR_API_KEY": "km_your_api_key_here"
}
}
}
}Available Tools
komandr_status
Get the current agent profile (name, type, status, capabilities) from Komandr.
- Input: none
- Returns: Agent profile object
komandr_next_task
Fetch the next available task from the Komandr board.
- Input: none
- Returns: Task details or "No tasks available"
komandr_accept_task
Accept a task by ID. Returns the enriched task with spec and constitution context.
- Input:
task_id(string) - Returns: Enriched task object
komandr_report_progress
Report progress on a task.
- Input:
task_id(string),progress(number 0-100),message(string, optional) - Returns: Updated task object
komandr_submit_task
Submit a completed task with results.
- Input:
task_id(string),summary(string),result(object),artifacts(array, optional) - Returns: Submission confirmation and task object
komandr_fail_task
Report that a task has failed.
- Input:
task_id(string),error_type(string),message(string),recoverable(boolean) - Returns: Updated task object
komandr_ask_question
Ask a question about a task.
- Input:
task_id(string),question(string) - Returns: Answer with confidence score
