list_tasksLABSCO SUMMARY
A backlog with real parent-child structure, reachable from the agent and from a browser.
The hierarchy is the point — create_task takes a parent id, and get_task_hierarchy returns the tree, so epics and their subtasks stay related instead of flattening into a list. The same store is visible in the web UI, so a human can look at what the agent did.
What it is
A Go workflow tracker with a web UI and a REST API, plus an MCP server that puts the same task store in front of an agent. Storage is either JSON files or PostgreSQL.
What you get
- list_tasks — filter by status and priority
- create_task — title, description, status, priority, and a parent for hierarchy
- get_task, update_task, delete_task
- get_task_hierarchy — the whole tree with parent-child relationships
- Three read-only resources: tasks://all, tasks://hierarchy, and tasks://summary with counts by status and priority
Requirements
Go 1.24 or later, or the container image. The MCP server takes MCP_PORT (default 3001) and STORAGE_DIR (default ./data). The web app and chat interface are separate and need their own LLM provider settings; the MCP path does not.
Setup effort
Build from source — git clone https://github.com/aykay76/projectflow.git
USE CASES
01
Breaking work down while you plan it
Create the epic, then create subtasks under it by passing the parent id, and read the tree back to check the shape.
02
Asking what is actually in progress
Filter the list by status and priority instead of reading the whole backlog.
03
Closing out work the agent just finished
Update the status and description on the task it was working from, so the record matches the code.
04
Getting a project summary without a query
Read the summary resource for total count, breakdown by status and priority, and recent activity.
6 tools
Listed in project docs. Your client may show a different list.
create_taskget_taskupdate_taskdelete_taskget_task_hierarchyBEFORE YOU INSTALL
What to know first
Tools you get
6 tools named in the project's documentation.
Requires
- go on PATH
- Go 1.24 or later
INSTALLATION
Connect your client
Server setup
TERMINAL
git clone https://github.com/aykay76/projectflow.git
cd projectflow
go build -o bin/mcp-server cmd/mcp-server/main.goMCP Clients
Claude Desktop, Cline
JSON
{
"mcpServers": {
"projectflow": {
"command": "<YOUR_PATH_TO_PROJECTFLOW>/bin/mcp-server",
"args": [],
"env": {
"STORAGE_DIR": "<YOUR_PATH_TO_PROJECTFLOW>/data"
}
}
}
}<YOUR_PATH_TO_PROJECTFLOW> is the folder you cloned. bin/mcp-server is the binary the go build step made. The data folder is given as a full path because the client does not start the server inside the clone.
Continue
CONTINUE · JSON
{
"mcpServers": [
{
"name": "projectflow",
"command": "<YOUR_PATH_TO_PROJECTFLOW>/bin/mcp-server",
"args": [],
"env": {
"STORAGE_DIR": "<YOUR_PATH_TO_PROJECTFLOW>/data"
}
}
]
}<YOUR_PATH_TO_PROJECTFLOW> is the folder you cloned. bin/mcp-server is the binary the go build step made. The data folder is given as a full path because the client does not start the server inside the clone.
VS Code
VS CODE · JSON
{
"servers": {
"projectflow": {
"command": "<YOUR_PATH_TO_PROJECTFLOW>/bin/mcp-server",
"args": [],
"env": {
"STORAGE_DIR": "<YOUR_PATH_TO_PROJECTFLOW>/data"
}
}
}
}<YOUR_PATH_TO_PROJECTFLOW> is the folder you cloned. bin/mcp-server is the binary the go build step made. The data folder is given as a full path because the client does not start the server inside the clone.
