AI Agents

Safely deploy AI to investigate, validate, escalate, or close Issues, governed by robust, user-defined Policies.

---

Overview

Method’s Agents framework enables safe deployment of AI agents with granular governance. Agents can investigate Issues, validate findings, escalate or close them, and draft Reports capturing their reasoning. They operate using MCP Tools that ship with the platform, and their behavior is controlled by Policies that enforce permissioned and transparent operation.

Agent components

A Method Agent is composed of:

• System prompt: Defines the agent’s role and behavior
• Model: The AI model the agent uses for reasoning
• Targets: The environments or objects the agent operates on
• MCP Tools: The set of tools the agent is authorized to use

Together, these define what the agent can do. Policies determine where and under what conditions it is allowed to do it.

The Agent Fleet application

The Agent Fleet application is where you create, view, and manage your Agents. Use it to configure new agents, attach MCP Tools, and adjust system prompts and models. Combine this with Policies to enforce approve, deny, or require-approval rules for agent actions across Agents, Environments, MCP Tools, and Agent Sessions.

Method-built Agent Fleet

Method ships with pre-configured agents designed for common workflows. For example, investigating and triaging Issues or drafting Reports. These can be used as-is or duplicated and customized to match your processes.

MCP Tools

The MCP Tools tab in the Tools app lets you browse and explore the tools available for agents. Each tool includes a description of its purpose and parameters, so you can choose the right set of capabilities when configuring an agent.

---

Policies

Policies govern AI agent behavior with granular approve, deny, and require-approval rules. They work alongside Agents: an Agent defines what the agent is; Policies define where and under what conditions those actions are allowed. These rules determine whether an agent can take an action automatically, is blocked, or must wait for human approval before proceeding.

Policy scope

Policies can be scoped to one or more of:

• Agent: A specific Agent. Use this to allow or restrict particular agents.
• Environment: A specific Environment. Use this to restrict agent actions in certain environments (e.g. production vs. development).
• MCP Tool: A specific tool the agent can call. Use this to gate individual tool usage.
• Agent Session: A single run or chat session. Use this for session-specific rules rather than agent-wide rules.

Scopes can be combined to express rules such as “this agent may use this tool in this environment only with approval” or “this agent is denied in production.”

Policy effects

Each policy has one of three effects:

• Approve: The action is allowed without further checks.
• Deny: The action is blocked.
• Require approval: The action is paused until a user approves it.

When multiple policies apply, deny overrides approve, and require-approval can gate sensitive actions.

---