Which Tools Does the Finance Plugin Integrate With? Complete MCP Connector Guide
The Finance plugin integrates with BigQuery, Microsoft 365, and Slack via pre-configured MCP servers, while remaining tool-agnostic through category placeholders that support Snowflake, Teams, NetSuite, and additional systems.
The Finance plugin in the anthropics/knowledge-work-plugins repository uses a tool-agnostic architecture that decouples skill logic from specific vendor implementations. Instead of hardcoding integrations, it defines workflow categories like data warehouse and chat, then binds these to MCP (Meta-Control-Plane) servers at runtime. This design lets finance teams swap tools without rewriting skills.
How the Finance Plugin Handles Tool Integration
The plugin achieves flexibility through category-based placeholders. When you write a skill file, you use generic tags like ~~data warehouse or ~~email instead of vendor-specific API calls. At runtime, the plugin resolves these placeholders to concrete MCP servers defined in finance/.mcp.json.
This architecture means the Finance plugin does not lock you into specific vendors. If your organization switches from BigQuery to Snowflake, you update the MCP server configuration without touching the skill files that reference ~~data warehouse.
Currently Configured MCP Servers
Out of the box, the Finance plugin ships with three active MCP server configurations in .mcp.json:
BigQuery (Data Warehouse)
The primary data warehouse integration points to Google BigQuery:
{
"server": "https://bigquery.googleapis.com/mcp",
"category": "data warehouse"
}
This server handles SQL queries against financial datasets. According to finance/CONNECTORS.md, BigQuery is the only pre-configured warehouse, though the placeholder system supports others.
Microsoft 365 (Email and Office Suite)
Microsoft 365 handles both email and office suite categories through a single MCP endpoint:
{
"server": "https://microsoft365.mcp.claude.com/mcp",
"categories": ["email", "office suite"]
}
This integration enables sending variance reports via Outlook and generating Word documents or Excel spreadsheets from financial data.
Slack (Chat)
For team communication, the plugin configures Slack:
{
"server": "https://mcp.slack.com/mcp",
"category": "chat"
}
Finance teams use this to post automated variance alerts and forecast updates to channels like #finance.
Placeholder Categories for Future Integrations
The finance/CONNECTORS.md file documents additional categories that lack pre-configured servers but are ready for custom MCP bindings:
ERP and Accounting Systems
The ~~erp placeholder supports:
- NetSuite
- SAP
- QuickBooks
- Xero
Analytics and BI Platforms
The ~~analytics category awaits configuration for:
- Tableau
- Looker
- Power BI
Alternative Data Warehouses
While BigQuery is the default, the ~~data warehouse placeholder also recognizes:
- Snowflake
- Databricks
- Redshift
- PostgreSQL
Referencing Tools in Skill Files
Skills use YAML frontmatter and markdown content with placeholder syntax. Here are concrete implementations from the repository:
Querying BigQuery via Placeholder
From the variance analysis skill, this pattern pulls data without naming BigQuery specifically:
---
name: revenue-forecast
description: Pull historical revenue data and generate a forecast.
---
# Revenue Forecast
Fetch the last 12 months of revenue from the data warehouse:
```sql
SELECT date, revenue FROM finance.revenue
WHERE date >= DATE_SUB(CURRENT_DATE(), INTERVAL 12 MONTH)
using the ~~data warehouse connector.
When executed, `~~data warehouse` resolves to `https://bigquery.googleapis.com/mcp` based on [`.mcp.json`](https://github.com/anthropics/knowledge-work-plugins/blob/main/.mcp.json).
### Posting to Slack
This skill sends variance notifications through the chat category:
```yaml
---
name: variance-notify
description: Post variance insights to the finance channel.
---
# Variance Notification
Generate the variance narrative and then:
```http
POST https://mcp.slack.com/mcp/api/chat.postMessage
{
"channel": "#finance",
"text": "{{variance_narrative}}"
}
using the ~~chat connector.
### Emailing via Microsoft 365
For executive reporting, this skill uses the email placeholder:
```yaml
---
name: variance-email
description: Email the CFO a PDF of the variance report.
---
# Email Variance Report
Render the report as PDF → attach → send via `~~email`:
```http
POST https://microsoft365.mcp.claude.com/mcp/api/mail.send
{
"to": "[email protected]",
"subject": "Monthly Variance Report",
"attachments": ["report.pdf"]
}
## Key Configuration Files
The Finance plugin defines tool integrations in two primary locations:
- **[`finance/.mcp.json`](https://github.com/anthropics/knowledge-work-plugins/blob/main/finance/.mcp.json)**: Contains concrete MCP server URLs, OAuth configurations, and category mappings. This file determines which integrations are active today.
- **[`finance/CONNECTORS.md`](https://github.com/anthropics/knowledge-work-plugins/blob/main/finance/CONNECTORS.md)**: Documents the complete matrix of supported categories, placeholders like `~~erp`, and future integration options such as Microsoft Teams or Snowflake.
## Summary
- The Finance plugin uses **tool-agnostic placeholders** (`~~data warehouse`, `~~email`) to avoid vendor lock-in.
- Currently configured MCP servers include **BigQuery**, **Microsoft 365**, and **Slack** as defined in [`.mcp.json`](https://github.com/anthropics/knowledge-work-plugins/blob/main/.mcp.json).
- Additional tools like **Snowflake**, **NetSuite**, and **Teams** are supported via placeholders but require manual MCP server configuration.
- Skills reference categories, not specific APIs, making tool migrations transparent to business logic.
- Configuration and supported options are documented in [`finance/CONNECTORS.md`](https://github.com/anthropics/knowledge-work-plugins/blob/main/finance/CONNECTORS.md) and [`finance/.mcp.json`](https://github.com/anthropics/knowledge-work-plugins/blob/main/finance/.mcp.json).
## Frequently Asked Questions
### What is MCP in the Finance plugin?
MCP stands for Meta-Control-Plane. It is a protocol that standardizes how the Finance plugin communicates with external tools. Instead of embedding API logic directly in skills, the plugin routes requests through MCP servers that translate generic category commands (like `~~chat`) into vendor-specific API calls.
### Can I use Snowflake instead of BigQuery?
Yes. While the Finance plugin ships with BigQuery pre-configured in [`.mcp.json`](https://github.com/anthropics/knowledge-work-plugins/blob/main/.mcp.json), the `~~data warehouse` placeholder supports Snowflake. You would configure a Snowflake MCP server in [`.mcp.json`](https://github.com/anthropics/knowledge-work-plugins/blob/main/.mcp.json), and existing skills using `~~data warehouse` would automatically route to Snowflake without code changes.
### How do I add a new tool integration?
To add a new integration, define an MCP server in [`finance/.mcp.json`](https://github.com/anthropics/knowledge-work-plugins/blob/main/finance/.mcp.json) with the appropriate category tag. For example, to add NetSuite under ERP, you would create an entry mapping `https://netsuite.example.com/mcp` to the `~~erp` category. The plugin will then route any skill using `~~erp` to your NetSuite server.
### Is Microsoft Teams supported?
Microsoft Teams is listed as a supported option for the `~~chat` category in [`finance/CONNECTORS.md`](https://github.com/anthropics/knowledge-work-plugins/blob/main/finance/CONNECTORS.md), but it is not pre-configured in [`.mcp.json`](https://github.com/anthropics/knowledge-work-plugins/blob/main/.mcp.json). You must add a Teams MCP server manually to enable it, similar to how Slack is currently configured.
Have a question about this repo?
These articles cover the highlights, but your codebase questions are specific. Give your agent direct access to the source. Share this with your agent to get started:
curl -s "https://instagit.com/install.md" Maintain an open-source project? Get it listed too →