Agents

The Agents area is the reusable behavior layer for Mogplex.

The top-level /agents route redirects to /agents/roster, and the product sub-nav is split into:

• /agents/roster
• /agents/skills
• /agents/rules
• /agents/context

What lives here

• Roster at /agents/roster for browsing built-in agents, creating custom agents, forking presets, and editing the model, prompt, description, and category for each one.
• Skills for reusable skill definitions the agent can load into a task.
• Rules for shared instruction blocks and policy-like guidance.
• Context for reusable context snippets that can be injected into sessions.

What you can do

• Keep a stable roster of presets and custom agents
• Group agents into built-in or custom categories
• Swap models without rebuilding the rest of the agent definition
• Reuse the same agent across triggers, assignments, and automations
• Build a shared library of skills, rules, and context that the web app and CLI can both reference

What the roster shows

The roster is not just a static list of prompts.

It merges built-in preset agents with your owned agents and forks, then shows live operating signals such as:

• running, idle, or error state
• cycle count
• last activity
• recent run and failure summary
• token usage summary when recent calls exist

That makes the roster useful both for editing definitions and for checking whether a reusable agent is actually being exercised in the product.

Important distinction

The agent definition is the reusable base. Flow nodes inside Automations can override some behavior, such as model or system prompt, without changing the master agent.

Assignments and triggers then reference those reusable agents later. In other words:

• define the reusable behavior in Agents
• bind it to a repo in Assignments
• bind it to a GitHub event in Triggers
• orchestrate multiple steps around it in Automations

If you want to route agent work to a repo, start from Projects, Triggers, or Automations.