The Fieldguide MCP lets you connect Fieldguide to an MCP-compatible AI tool. After you connect it, you can ask your AI tool to retrieve Fieldguide data or complete supported Fieldguide actions without opening Fieldguide for each step.
⠀
MCP stands for Model Context Protocol. It is a standard that lets an AI tool connect to other systems, such as Fieldguide, through a defined set of tools and permissions.
Choose your connection method
The Fieldguide MCP supports two connection methods. To choose one, match your AI tool to the method:
OAuth (beta): Use this method for Claude Enterprise and Microsoft Copilot Studio. A Fieldguide System Admin enables the integration in Fieldguide, and the AI tool connects with OAuth instead of an API token. Follow the article for your tool:
API token: Use this method for other MCP-compatible AI tools that support token-based authentication. Follow the steps in the rest of this article, starting with Generate a Fieldguide API token.
⠀
Generate a Fieldguide API token
If you connect your AI tool with an API token instead of OAuth, the token authenticates your AI tool to Fieldguide.
Before you set up the Fieldguide MCP with an API token, make sure you have:
A Fieldguide account.
Access to the Fieldguide data you want to use through the MCP.
Permission to create a Fieldguide API token.
An MCP-compatible AI tool that supports token-based authentication.
The Fieldguide MCP uses your Fieldguide permissions. If you don't have access to an engagement, request, or action in Fieldguide, your AI tool can't access it through the MCP.
⠀
To create a token, follow the Create an API token steps in Fieldguide API.
⠀
Store the token somewhere secure after you create it. Fieldguide only shows the token once.
⠀
Connect your AI tool with an API token
The setup flow depends on your AI tool. In the tool's MCP connection settings, create a new connection using these values:
Server URL:
https://api.fieldguide.io/mcpAuthentication type: API key or bearer token authentication
Authorization header:
Authorization: Bearer YOUR_TOKEN
⠀
Replace YOUR_TOKEN with the Fieldguide API token you generated.
⠀
Connecting the MCP doesn't move Fieldguide data into your AI tool on its own. Your AI tool sends requests to Fieldguide only when you ask it to use the connection.
⠀
What the Fieldguide MCP supports today
The Fieldguide MCP supports conversational access to common engagement and request workflows. The exact response depends on your Fieldguide permissions and the details you include in your prompt.
⠀
Review engagement status
Ask for a status summary when you want to understand where an engagement stands.
⠀
Example prompts:
What's the current status of the [ENGAGEMENT NAME] engagement?Which requests are open or overdue on [ENGAGEMENT NAME]?Who owns the overdue requests for [ENGAGEMENT NAME]?
⠀
The AI tool can retrieve engagement details and request information, then summarize open requests, overdue requests, owners, and status.
⠀
List outstanding requests
Ask for open or overdue requests when you need a request list across one or more engagements.
⠀
Example prompts:
Show me open requests for [CLIENT NAME].Which requests are overdue across my active engagements?Show me the open requests for [ENGAGEMENT NAME], grouped by owner.
⠀
For broad requests, include a client, engagement, owner, or due-date range in your prompt to help your AI tool return the right results.
⠀
Post follow-up comments on requests
Ask your AI tool to draft and post follow-up comments on Fieldguide requests.
⠀
Example prompts:
Post a follow-up comment on the overdue requests for [CLIENT NAME].Add this comment to [REQUEST NAME]: [COMMENT TEXT].Draft a reminder comment for requests due this week on [ENGAGEMENT NAME].
⠀
For write actions, review the request list and comment text before your AI tool posts anything in Fieldguide. This keeps you in control of what gets added to request activity.
⠀
Create an engagement from a template
Ask your AI tool to create a new engagement from an existing engagement template.
⠀
Example prompts:
Create a new engagement for [COMPANY NAME] from our standard [TEMPLATE NAME] template.Set up a [PERIOD] engagement for [COMPANY NAME] using [TEMPLATE NAME].
⠀
If required details are missing, your AI tool should ask for them before creating the engagement. These details may include the company, template, engagement name, dates, or report issue date.
⠀
What the Fieldguide MCP doesn't support yet
The Fieldguide MCP currently focuses on engagement and request workflows. It doesn't support:
OAuth-based MCP setup for AI tools other than Claude Enterprise and Microsoft Copilot Studio.
Insights queries.
Triggering Field Agents, agent runs, or Orchestrator workflows.
Directly modifying documents, forms, or sheets.
⠀
Scopes
A scope is a permission that controls what your AI tool can access in Fieldguide. You select scopes when you enable the Fieldguide MCP for an OAuth connection or when you create an API token.
Scopes come in two types:
Read scopes: Let your AI tool view Fieldguide data. An example is
engagements:read.Write scopes: Let your AI tool change Fieldguide data, such as posting comments or creating engagements. Examples are
comments:writeandengagements:write.
Select the narrowest set of scopes that covers what you want to do. For example, read scopes are enough to review engagement status and list open requests. You need write scopes to post comments or create engagements. If you want your AI tool to only retrieve data, don't select write scopes.
Security and access
The Fieldguide MCP acts on your behalf, using either an OAuth connection or your Fieldguide API token. The connection's access is limited by:
Your Fieldguide user permissions.
The scopes you select when you enable the integration or create the API token.
Any Fieldguide access restrictions that apply to the underlying engagement, request, or object.
⠀
If you use an API token, keep it secure. Don't paste it into prompts or share it in messages. Add it only to your AI tool's MCP connection settings.
⠀
If your AI tool can't access data you expect it to find, check that your Fieldguide account has access to that data and that the scopes you selected include it.
⠀