backlog-mcp

A task backlog MCP server for LLM agents. Works with any MCP client — Claude, Kiro, Cursor, Codex, etc.

Agents create tasks, track progress, attach artifacts, and search across everything. Humans get a real-time web viewer to see what agents are doing.

Runs locally out of the box. Can also be self-hosted on Cloudflare Workers + D1 for a free, always-on remote backlog accessible from any device or MCP client.

Quick start: Tell your LLM: Add backlog-mcp to .mcp.json and use it to track tasks

Live demo: backlog-mcp-viewer.pages.dev — the viewer UI connected to a real hosted instance

What's Inside

This is a monorepo with 3 packages:

The viewer is built with Nisli, a zero-dependency reactive Web Component framework published as @nisli/core. Nisli started in this repo and now lives separately.

Installation

Add to your MCP config (.mcp.json or your MCP client config):

{
  "mcpServers": {
    "backlog": {
      "command": "npx",
      "args": ["-y", "backlog-mcp"]
    }
  }
}

Self-Hosting on Cloudflare (Optional)

Host your own always-on remote backlog for free using Cloudflare Workers + D1. Accessible from any device or MCP client — no local server process required.

Prerequisites: Cloudflare account (free tier is enough), wrangler CLI, a GitHub OAuth App.

# 1. Clone and install
git clone https://github.com/gkoreli/backlog-mcp.git
cd backlog-mcp
pnpm install

# 2. Create the D1 database
cd packages/server
npx wrangler d1 create backlog
# Copy the database_id into packages/server/wrangler.jsonc

# 3. Apply the D1 migrations
npx wrangler d1 execute backlog --remote --file=migrations/0001_initial.sql
npx wrangler d1 execute backlog --remote --file=migrations/0002_oauth_refresh_tokens.sql

# 4. Set secrets
npx wrangler secret put JWT_SECRET          # any strong random string
npx wrangler secret put API_KEY             # your personal API key (for Claude Desktop)
npx wrangler secret put GITHUB_CLIENT_ID    # GitHub OAuth App client ID
npx wrangler secret put GITHUB_CLIENT_SECRET
npx wrangler secret put ALLOWED_GITHUB_USERNAMES  # comma-separated: "you,youralt"

# 5. Deploy
npx wrangler deploy

GitHub OAuth App setup: go to GitHub → Settings → Developer settings → OAuth Apps → New. Set the callback URL to https://<your-worker>.workers.dev/oauth/github/callback.

Once deployed, add to your MCP client config:

{
  "mcpServers": {
    "backlog": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://<your-worker>.workers.dev/mcp"]
    }
  }
}

Claude.ai and ChatGPT can connect directly via the remote MCP URL — no local process needed. GitHub OAuth sessions use rotating refresh tokens so clients can renew access without daily re-authentication.

Web Viewer

Open http://localhost:3030 — always available when the server is running.

Features:

• Split pane layout with task list and detail view
• Spotlight search with hybrid text + semantic matching
• Real-time updates via SSE
• Activity timeline
• Filter by status, type, epic
• GitHub-style markdown rendering with Mermaid diagrams
• URL state persistence

The viewer UI is built with Nisli (@nisli/core).

Entity Types

5 entity types, all stored as markdown files with YAML frontmatter:

Status values: open, in_progress, blocked, done, cancelled

Example task file:

---
id: TASK-0001
title: Fix authentication flow
status: open
epic_id: EPIC-0002
parent_id: FLDR-0001
references:
  - url: https://github.com/org/repo/issues/123
    title: Related issue
evidence:
  - Fixed in PR #45
---

The authentication flow has an issue where...

MCP Tools

backlog_list

backlog_list                              # Active tasks (open, in_progress, blocked)
backlog_list status=["done"]              # Completed tasks
backlog_list type="epic"                  # Only epics
backlog_list epic_id="EPIC-0002"          # Tasks in an epic
backlog_list parent_id="FLDR-0001"        # Items in a folder
backlog_list query="authentication"       # Search across all fields
backlog_list counts=true                  # Include counts by status/type
backlog_list limit=50                     # Limit results

backlog_get

backlog_get id="TASK-0001"                # Single item
backlog_get id=["TASK-0001","EPIC-0002"]  # Batch get

backlog_create

backlog_create title="Fix bug"
backlog_create title="Fix bug" description="Details..." epic_id="EPIC-0002"
backlog_create title="Q1 Goals" type="epic"
backlog_create title="Research notes" type="artifact" parent_id="TASK-0001"
backlog_create title="v2.0 Release" type="milestone" due_date="2026-03-01"
backlog_create title="Fix bug" source_path="/path/to/spec.md"  # Read description from file

backlog_update

backlog_update id="TASK-0001" status="done"
backlog_update id="TASK-0001" status="blocked" blocked_reason=["Waiting on API"]
backlog_update id="TASK-0001" evidence=["Fixed in PR #45"]
backlog_update id="TASK-0001" parent_id="FLDR-0001"
backlog_update id="MLST-0001" due_date="2026-04-01"

backlog_delete

backlog_delete id="TASK-0001"             # Permanent delete

backlog_search

Full-text + semantic hybrid search with relevance scoring:

backlog_search query="authentication bug"
backlog_search query="design decisions" types=["artifact"]
backlog_search query="blocked tasks" status=["blocked"] limit=10
backlog_search query="framework" sort="recent"
backlog_search query="search ranking" include_content=true

backlog_context

Get rich context for a task — parent epic, siblings, children, cross-references, reverse references, recent activity, and semantically related items:

backlog_context task_id="TASK-0001"
backlog_context task_id="TASK-0001" depth=2          # Grandparent/grandchildren
backlog_context query="search ranking improvements"   # Find by description
backlog_context task_id="TASK-0001" include_related=false  # Skip semantic search

write_resource

Edit existing files on the MCP server. All creation goes through backlog_create.

# Edit task body (use str_replace — protects frontmatter)
write_resource uri="mcp://backlog/tasks/TASK-0001.md" \
  operation={type: "str_replace", old_str: "old text", new_str: "new text"}

# Insert after a specific line
write_resource uri="mcp://backlog/tasks/TASK-0001.md" \
  operation={type: "insert", insert_line: 5, new_str: "inserted line"}

# Append to a file
write_resource uri="mcp://backlog/resources/log.md" \
  operation={type: "append", new_str: "New entry"}

Operations: str_replace (exact match, must be unique), insert (after line number), append (end of file).

How It Works

Running npx -y backlog-mcp (the default MCP config) does the following:

1. Starts a persistent HTTP server as a detached background process — serves both the MCP endpoint (/mcp) and the web viewer (/) on port 3030
2. Bridges stdio to it — your MCP client communicates via stdio, which gets forwarded to the HTTP server via mcp-remote
3. Auto-updates: npx -y always pulls the latest published version. If the running server is an older version, it's automatically shut down and restarted with the new one
4. Resilient recovery: If the bridge loses connection, a supervisor restarts it with exponential backoff (up to 10 retries). Connection errors like ECONNREFUSED are detected and handled automatically

The HTTP server persists across agent sessions — multiple MCP clients can share it. The web viewer is always available at http://localhost:3030.

CLI

All commands via npx:

npx backlog-mcp                # Start stdio bridge + auto-spawn HTTP server (default)
npx backlog-mcp status         # Check server status
npx backlog-mcp stop           # Stop the server
npx backlog-mcp version        # Show version
npx backlog-mcp serve          # Run HTTP server in foreground (optional, see below)

Sample outputs:

$ npx backlog-mcp status
Server is running on port 3030
Version: 0.44.0
Data directory: /Users/you/.backlog
Task count: 451
Uptime: 3515s
Viewer: http://localhost:3030/
MCP endpoint: http://localhost:3030/mcp

$ npx backlog-mcp stop
Stopping server on port 3030...
Server stopped

$ npx backlog-mcp status
Server is not running

The CLI exists for humans to inspect and manage the background server that agents use. Since the default mode spawns a detached process, you need status to check it and stop to shut it down.

serve runs the HTTP server in the foreground instead of detached — useful for debugging, Docker containers, or running without an MCP client. In normal usage you never need it; the default command handles everything.

Configuration

BACKLOG_DATA_DIR=~/.backlog    # Where to store tasks (default: data/tasks/)
BACKLOG_VIEWER_PORT=3030       # HTTP server port

Create a .env file for local development — see .env.example.

Development

git clone https://github.com/gkoreli/backlog-mcp.git
cd backlog-mcp
pnpm install
pnpm build          # Build all packages
pnpm test           # Run all workspace tests
pnpm dev            # Server + viewer with hot reload

Architecture

packages/
├── server/       # MCP server, search, context hydration, storage
├── viewer/       # Web UI built with @nisli/core
└── shared/       # Entity types, ID utilities
docs/
├── adr/              # backlog-mcp architecture decision records
└── framework-adr/    # Pointer to Nisli ADRs

Backlog ADRs document significant design decisions. See docs/adr/README.md for the full index. Nisli ADRs live in the Nisli repository.

License

MIT