Context Weaver
An MCP server that gives LLMs read-only access to Microsoft Project Online (Project for the Web) via the Microsoft Graph API.
README
Context Weaver
An MCP server that gives LLMs read-only access to Microsoft Project Online (Project for the Web) via the Microsoft Graph API.
Built on the Model Context Protocol — works with Claude Desktop, Claude Code, and any MCP-compatible client.
What It Does
Context Weaver exposes 7 read-only tools that let an LLM query your Microsoft Project data:
| Tool | Description |
|---|---|
authenticate |
Connect your Microsoft 365 account (OAuth 2.0 + PKCE) |
list_projects |
List all plans you have access to |
get_project |
Get project details, buckets, and task count |
list_tasks |
List tasks with filters (bucket, assignee, status) |
get_task |
Full task detail including description and checklist |
list_buckets |
List project phases/columns |
get_assignments |
Your tasks across all projects |
check_overdue |
Find overdue incomplete tasks |
v0 is read-only by design. No task mutations, no accidental side effects.
Prerequisites
- Node.js ≥ 20
- Microsoft 365 account with Project for the Web (Planner-based)
- Azure AD app registration (see Setup below)
Setup
1. Register an Azure AD App
- Go to Azure Portal → App registrations
- Click New registration
- Name:
Context Weaver - Supported account types: Single tenant (or multi-tenant if needed)
- Redirect URI: Web →
http://localhost:3847/auth/callback
- Name:
- Under Authentication, enable Allow public client flows
- Under API permissions, add:
Tasks.Read(Delegated)Tasks.Read.Shared(Delegated)Group.Read.All(Delegated)User.Read(Delegated)
- Copy the Application (client) ID and Directory (tenant) ID
2. Configure Environment
cp .env.example .env
Edit .env:
AZURE_CLIENT_ID=your-client-id
AZURE_TENANT_ID=your-tenant-id
3. Install & Build
npm install
npm run build
Usage
With Claude Desktop
Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
{
"mcpServers": {
"context-weaver": {
"command": "node",
"args": ["/absolute/path/to/context-weaver/dist/index.js"],
"env": {
"AZURE_CLIENT_ID": "your-client-id",
"AZURE_TENANT_ID": "your-tenant-id"
}
}
}
}
With Claude Code
claude mcp add context-weaver node /absolute/path/to/context-weaver/dist/index.js
Development
npm run dev
Example Prompts
Once connected, try:
- "What projects do I have access to?"
- "Show me all overdue tasks"
- "List the tasks in the Design phase of Project X"
- "What's assigned to me this week?"
- "Get the full details and checklist for task ABC123"
Architecture
LLM Client ◄──── MCP (stdio) ────► Context Weaver ──── Graph API ────► Project Online
(TypeScript) (OAuth 2.0) (Planner)
Project Structure
src/
├── index.ts # MCP server entry, tool registration
├── auth/
│ ├── oauth.ts # PKCE auth flow
│ └── session.ts # In-memory token store
├── graph/
│ ├── client.ts # Graph API HTTP client w/ retry
│ ├── projects.ts # Plan/bucket queries
│ ├── tasks.ts # Task queries
│ └── users.ts # User name resolution + cache
├── tools/
│ ├── list-projects.ts
│ ├── get-project.ts
│ ├── list-tasks.ts
│ ├── get-task.ts
│ ├── list-buckets.ts
│ ├── get-assignments.ts
│ └── check-overdue.ts
└── utils/
├── errors.ts # Error mapping + priority labels
└── types.ts # Shared TypeScript interfaces
Design Decisions
- Project for the Web (Planner-based), not classic Project Online (PWA/OData). Planner has better Graph API coverage and is Microsoft's forward-looking product.
- Per-user delegated auth — the server never holds elevated service account privileges.
- Read-only in v0 — write operations planned for v1 with human-in-the-loop confirmation.
- Three dependencies —
@modelcontextprotocol/sdk,@azure/msal-node,zod. Graph calls use nativefetch.
Roadmap
- [ ] v0.1 — Read-only MVP (this release)
- [ ] v0.2 — Write operations (create/update tasks) with confirmation prompts
- [ ] v0.3 — Dependency tracking (Dataverse or local graph)
- [ ] v0.4 — Slack connector (search decisions, summarize threads)
- [ ] v0.5 — GitHub connector (PR status, issue sync)
- [ ] v1.0 — Self-correcting agent loop (schedule validation)
Known Limitations
- Planner doesn't support task dependencies natively. Dependency tracking requires Dataverse direct access or a local graph — planned for v0.3.
get_assignmentsonly works for the authenticated user. Fetching another user's assignments requires iterating all plans they're members of.- Tokens are in-memory only. You'll need to re-authenticate each session. Persistent encrypted storage planned for v0.2.
Contributing
PRs welcome. Please open an issue first for anything beyond a typo fix.
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.