Whatever message this page gives is out now! Go check it out!
Resource Concepts
Last update:
May 18, 2026
In MCP, a resource is any piece of data your tools and agents can read via a standard interface.
Overview
In MCP, a resource is any piece of data your tools and agents can read via a standard interface. Resources give the model context about your environment without requiring direct access to disks, databases, or APIs.
Typical examples:
- Log files, config files, or report outputs on disk
- Structured JSON indexes, catalogs, or metadata documents
- Remote documents fetched via HTTP or other protocols
- Virtual views backed by databases or services
Conceptually, a resource has:
- A URI – unique identifier (e.g.,
file:///var/log/app.logorlogs://today) - A name – human-friendly label (e.g., “Application Logs”)
- A MIME type – describes the data format (e.g.,
text/plain,application/json) - Optional metadata – tags, descriptions, or size hints
What You Can Do With Resources
From a user’s perspective, resources are read-only context:
- You can ask an agent to “look at the logs resource and summarize recent errors”.
- You can browse available resources via MCP (
resources/list). - You can fetch content via MCP (
resources/read) and let the model reason about it.
Key points:
- Resources are not tools; they don’t perform actions.
- They are not necessarily files; they can be virtual or computed on demand.
- They are safe views over underlying data, respecting whatever scoping and security your MCP server enforces.
Listing Resources
When you list resources, you’ll see items like:
{
"uri": "file:///var/app/logs/app.log",
"name": "Application Logs",
"mimeType": "text/plain"
}
or:
{
"uri": "reports://weekly/sales",
"name": "Weekly Sales Report",
"mimeType": "application/json"
}
You don’t need to know how they’re implemented; you just use the URI to read them.
We're glad. Tell us how this page helped.
We're sorry. Can you tell us what didn't work for you?
Share additional feedback to help us improve.
0/255
|
