Endpoint Agent overview

The QuilrAI Endpoint Agent discovers and governs AI use that happens outside the browser: desktop AI apps, coding agents, local models, and the MCP servers and skills they use. It runs as a background service on macOS and Windows and syncs what it finds to the QuilrAI console.

What it discovers and governs

Area	What the agent does
Desktop AI apps	Finds installed and running AI apps (for example ChatGPT, Claude, Copilot, Gemini) and applies per-app policies.
Coding agents	Discovers tools such as Cursor and Claude Code, plus their hooks, plugins, and configuration files.
Local models	Detects local model runtimes such as Ollama and LM Studio, and downloaded models.
MCP servers and skills	Inventories MCP server configurations and agent skills on the device.
Network monitoring	Inspects AI traffic with TLS inspection, runs DLP on requests and responses, and can block or ask for justification. See Network monitoring.
Process mapping	Maps running processes to known applications and enforces allow, block, quarantine, or justify policies. See Process mapping.

Where the data shows up

Console area	What you see
Inventory	Apps, models, MCP servers, skills, plugins, and hooks discovered on endpoints. See Inventory.
Agents	Agents observed on endpoints (sources Endpoint discovered and Endpoint runtime). See Agents.
Findings	DLP findings and enforcement outcomes from endpoint traffic. See Findings and interactions.
Users › Endpoint deployment	Which workstations run the agent, its version and status. See Deployment and status.
Skills Library › Discovered	Skills seen on endpoints, with apps, devices, scope, and risk. See Skills Library.

Get started

1. Check the requirements for your endpoints.
2. Get the agent packages from your QuilrAI representative and roll them out, following the installation guides at installdocs.quilrai.dev.
3. Review agent settings and set app policies.
4. Keep the kill switch procedure handy for incidents.

note

The Endpoint Agent is a separate package from the Browser Extension. Deploy both for full coverage of browser and desktop AI use.