Agents Configuration

Map AI clients to MCPs and monitor per-agent usage.

How It Works

Registered Agents

Cursor (cursor)

Claude (claude)

Custom: my-bot

MCP Mapping

Cursor → GitHub ✓ Slack ✓

Claude → GitHub ✓ S3 ✗

my-bot → Slack ✓

Usage Stats

Cursor: 1,247 calls

Claude: 892 calls

my-bot: 156 calls

QuilrAI

1. Register Agents - Use predefined agents or create custom ones
2. Map to MCPs - Enable or disable MCPs per agent
3. Monitor Usage - Track per-agent tool call statistics

Predefined Agents

Built-in agents are identified by their User-Agent header keywords:

Agent	User-Agent Keyword
OpenAI	openai
Claude	claude
Cursor	cursor
Gemini	gemini

Custom Agents

Create custom agents for any AI client not in the predefined list. Each custom agent requires:

• User-Agent Keyword - The keyword to match in the User-Agent header (e.g., my-custom-agent)
• Display Name - A human-readable name for the dashboard (e.g., My Custom Agent)

Per-Agent Dashboard

Each agent card shows:

• Total tool calls - Cumulative tool invocations for the agent
• MCP access - Which MCPs are enabled vs. disabled
• Toggle controls - Enable or disable individual MCP access per agent directly from this view

Example

Agent	Tool Calls	MCPs
OpenAI	1,247	GitHub, Slack, Jira, Confluence (enabled) - S3, Internal API (disabled)

Agents Configuration vs. Access Control

View	Best For
Agents Configuration	Global view - see all MCPs for each agent. Best for managing agent permissions across your entire MCP fleet.
Access Control	Per-MCP view - see all agents for one MCP. Best when configuring a single MCP's permissions.

Both views control the same underlying permissions - use whichever is more convenient for your workflow.