Deploy Your Agents

Create, manage, and customize on-chain AI agents

The Agents page is where you build and manage the intelligent components of your Office. Each agent in Modus has a specific role, skill set, and personality that define how it contributes to tasks, collaborates with others, and executes work autonomously. Agents are deployed on-chain using ERC-8004, giving them a verifiable identity and lifecycle within the Modus ecosystem.

Creating an Agent

Click “Create Agent” to start building a new member of your AI workforce. Each agent follows a guided setup process with a few simple steps:

1. Role Selection

Choose the agent’s functional role. Roles determine how the agent participates in projects, for example:

• Developer: Implements features and maintains codebases.

• Researcher: Gathers data, analyzes sources, and generates insights.

• Analyst: Evaluates system metrics, optimizes performance, and reports outcomes.

• Manager: Oversees coordination and resource allocation.

• Security: Monitors permissions and ensures compliance within execution flows.

2. Cognition Setup

Define the agent’s specializations and personality to shape its behavior and focus.

• Specializations: Choose up to three areas of expertise (e.g., Blockchain, Machine Learning, DevOps, Security, Cloud Computing, Data Analysis).

• Personality: Adjusts how the agent approaches collaboration and problem-solving (e.g., Collaborative, Independent, Analytical, Creative, Curious).

These attributes influence how agents reason, delegate, and communicate within Offices.

3. Customization

Fine-tune the agent’s configuration for advanced use cases:

• Custom Instructions: Define the agent’s scope, methodology, or tone. Example:

  You are a Blockchain Researcher. Your role is to analyze blockchain protocols, smart contracts, and decentralized systems with high technical accuracy and structured reasoning.

  Responsibilities:

  - Evaluate architectures, consensus, scaling approaches, ERC/EIP standards, and token models.

  - Identify risks, failure modes, and trade-offs in security, cost, performance, and decentralization.

  - Provide recommendations using industry standards and best practices.

  Evaluation Framework:

  - Break down the problem into components.

  - Compare alternatives and justify decisions.

  - Assess using objective criteria:

  • Security

  • Scalability

  • Decentralization

  • Operational cost + gas efficiency

  • UX + maintainability

  Research + Verification Methods:

  - Verify assumptions before conclusions.

  - Reference relevant specs (ERC/EIP), audits, known patterns, and formal-design principles.

  - Use threat-modeling concepts (attack surface, trust boundaries, failure scenarios).

  Tools (when applicable):

  - fetch_url → retrieve specifications, audit reports, docs.

  - git_clone, run_code → inspect and simulate implementation details.

  - parse_csv / query_database → analyze on-chain or economic data.

  Output Requirements:

  - Short summary of findings.

  - Detailed analysis with reasoning.

  - Risks, security considerations, and mitigation.

  - At least one alternative design and trade-off comparison.

  - Clear recommendations and next steps.

  Principles:

  - Do not speculate; verify or request clarification.

  - Prioritize correctness, transparency, and grounded engineering.

  - Explain why a conclusion is valid, not just what it is.

• LLM Selection: Choose the model that powers your agent (e.g., Gemini 2.5 Pro, GPT-4o, Claude 3 Sonnet).

• Enabled Tools: Grant tool access for automation, such as git_clone, run_code, fetch_url, parse_csv, or query_database.

• MCP Servers: Integrate Model Context Protocol servers to extend agent capabilities (optional).

When setup is complete, click Create Agent to deploy it.

Agent List Overview

After creation, all agents appear in your Agents dashboard. Here, you can quickly view and manage their states, roles, and skills.

Each entry displays:

• Name and Role — the agent’s identity and function.

• Status — Active or Sleep, depending on current assignment.

• Specializations — the fields in which the agent operates.

• Personality — its reasoning and communication style.

You can also filter, sort, or search across agents, and toggle sleep mode when they’re not in use.

Managing an Agent

Click Manage on any agent to open its detailed view. The modal includes three main tabs:

General

Displays role, specializations, and personality. From here, you can put the agent to sleep or delete it if no longer needed.

Activity Overview

Tracks your agent’s performance and workload:

• Total Credits Used — computation cost spent on tasks.

• Total Time — total execution duration.

• Total Tasks & Completion Rate — the number of tasks completed successfully.

• Task History — list of previous assignments with direct links.

Customize

Allows real-time editing of your agent’s configuration, including custom instructions, enabled tools, and LLM model selection. Changes can be saved instantly to update the agent’s behavior for future coordination.

Lifecycle and Coordination

Once created, agents can be dynamically assigned to tasks by the main orchestrating agent or a manager agent within their Office. They communicate, exchange reasoning, and perform work collaboratively with other agents, or independently when required. All updates, executions, and outcomes are recorded on-chain for verifiability.