Skip to main content
Version: Cloud

What is possible with the MCP server

The Deeploy MCP server serves two different people, with two separate workflows. This page covers both.

For the exhaustive tool list, see Tools; for the skills that drive the agent, see Skills over MCP.

Two personas, two workflows​

Agent tracingUse case compliance
Who it is forAll usersCompliance officers
What it doesRegisters the agent and records every interaction as an audit trail.Reviews controls and assembles compliance evidence.
When it runsAutomatically, at the start of every session.Only when you explicitly ask for it.
Toolslimitedfull
Skilldeeploy-autoregisterdeeploy-use-case-controls

The distinction matters in daily use: tracing affects everyone on every session and needs no action from you, whereas compliance work never starts on its own — the agent will not mention it unless you ask.


Agent registration and tracing​

Session flow​

Every time an AI agent starts a session with the MCP server connected, the following happens automatically:

  1. The agent lists your existing registrations. It shows all agents already registered in your Deeploy Workspace.
  2. You choose a name. The agent asks what name to register under. You can use an existing name to resume tracing under that registration, or provide a new name to create one. Type skip to continue without registering or tracing.
  3. Every response is silently traced. Once registered, the agent sends an audit trail entry to Deeploy after every response for the rest of the session. This happens in the background — you won't see any confirmation unless something goes wrong.

This flow is driven by the server's instructions plus the .deeploy-autoregister skill. Sending that skill makes the flow more robust: it also triggers automatically rather than waiting for a manual prompt.

Agent registration​

Each agent is registered in Deeploy as a Deployment of type Registration. A registration stores:

  • The agent name and description
  • The source platform (e.g. Copilot Studio, Claude, GitHub Copilot)
  • The endpoint or URL of the agent, if available
  • Any custom metadata you want to attach (owner, team, version, capabilities, etc.)

Existing registrations can be used across sessions. Using the same name again adds new traces to the existing registration rather than creating a duplicate.

You can update a registration at any time by asking the agent to update it, or directly in the Deeploy interface.

Tracing​

Traces are visible in the Deployment detail view under Monitoring → Tracing. For each trace you can view:

  • The full input and output text
  • The interaction type
  • A timestamp
  • The session context (environment, model, user) attached to the root trace
  • Any custom metadata passed by the agent

Tools used​

Four tools, all called automatically: list_agents, register_agent, patch_agent, and send_trace. No user action is needed beyond choosing a registration name. See Tools → Agent registration and tracing.


For use case compliance​

This is the compliance-officer workflow: reviewing what a use case must demonstrate, and assembling the evidence for it.

What it does​

The agent helps with a use case's controls — the requirements the use case must satisfy, drawn from the control frameworks active on it. Working from your existing material, it can:

  • Read a use case's lifecycle stage to determine which controls apply.
  • Review the controls and the evidence already attached.
  • Add new evidence, or reference evidence another control already owns.
  • Attach documents and templates to checks that require one.
  • Fill in documentation templates.

It is manual and opt-in​

The agent engages only when you explicitly ask to work on use case controls or compliance evidence. It never starts this work on its own, and never mentions it unprompted — not all users are compliance officers. If you ask what the capability is, the agent answers briefly and waits to be asked.

Prompts that engage it:

  • "Help me fill in the controls for the fraud detection use case."
  • "What evidence is missing for this control?"
  • "Can our model card count as evidence for this control?"
  • "Review the AI Act controls for <use case>."

Evidence, not box-ticking​

The workflow is about demonstrating readiness, not filling in fields. The agent is instructed to assess critically whether the evidence genuinely supports the control, and to say so when it does not. A control with fields filled in is not automatically a control that has been satisfied.

Before creating or changing anything, the agent shows what it intends to write and asks for approval.

Where to start​

Ask the agent to list your use cases, then pick one. See Skills over MCP for how the workflow is guided, and Tools → Use case controls for the full tool list.