Skip to content

Tools overview

How an agent takes action and reads live data through MCP.


On this page

Tools let an agent do things instead of only describing them: look up a record, create a ticket, run a query, post to an API. Runbear connects them over MCP (Model Context Protocol), an open standard for giving agents access to external systems.

How a tool call works#

When a message needs something the agent does not already know, the agent picks a tool, calls the connected MCP server, and uses what comes back to answer. This happens mid-conversation, so the answer reflects the state of the system at that moment.

Asking "what is the status of RB-1234" makes the agent call the Linear tool, read the issue, and reply with its current status rather than a remembered one.

Managed servers and your own#

Managed catalogCustom MCP
HostingRunbearYou
SetupAuthorize with OAuthProvide a server URL
Good forCommon SaaS integrationsInternal tools and private APIs

Who the agent acts as#

A connected tool authorizes one of two ways.

Per-User means each member links their own account, and the agent acts with that member's permissions. Two people asking the same question can get different answers, which is usually what you want for a CRM or an issue tracker.

Shared means one account serves everyone on the agent. Simpler, but every user of the agent inherits that account's access, so grant it deliberately.

Tools or knowledge#

ToolKnowledge
SourceA live API callDocuments synced ahead of time
Answers"What is true right now" and "do this""What does our documentation say"
ExampleCreate a Linear issue, look up a paymentAnswer from the refund policy

Most agents use both.