mock-agent

Mock Agent Fleet Agent

A sophisticated mock agent that simulates realistic AI coding agent behavior for testing and demonstrating the Agent Fleet backend system.

Features

Core Functionality

• State Machine: Implements realistic agent states (idle, active, waiting_feedback, error)
• API Integration: Full integration with Agent Fleet backend REST API
• Command Processing: Handles instructions, feedback, and questions from the fleet management system
• Todo Management: Creates and manages realistic todo lists for tasks
• Progress Tracking: Simulates work progress with metrics (files changed, lines added/removed)
• Event Logging: Comprehensive logging of agent activities and state changes

Randomized Behavior Mode

• Dynamic State Changes: Random state transitions based on configurable probabilities
• Scenario Execution: Predefined realistic work scenarios with different complexity levels
• Error Simulation: Realistic error conditions and recovery patterns
• Question Generation: Context-aware questions requiring human feedback
• Metric Fluctuations: Realistic work progress simulation

Predefined Scenarios

• Bug Fix: Authentication issue resolution with testing and deployment
• Feature Development: User dashboard implementation with frontend/backend work
• Performance Optimization: Database query optimization with benchmarking
• Security Audit: API endpoint security review and vulnerability fixes
• Code Refactoring: Legacy code modernization with architectural improvements
• Infrastructure: CI/CD pipeline enhancement with automation
• Data Migration: Database schema updates with zero-downtime strategies
• Integration: Third-party service integration with error handling

Quick Start

Build and Run

# Build the mock agent
cd cmd/apps/agent-fleet/mock-agent
go build -o mock-agent .

# Run with default settings (connects to localhost:8080)
./mock-agent

# Run with randomized behavior
./mock-agent --randomized

# Run with custom configuration
./mock-agent \
  --server http://backend-server:8080 \
  --name "Advanced Code Reviewer" \
  --worktree /projects/my-app \
  --randomized \
  --tick-interval 3 \
  --log-level debug

Configuration Options

Usage:
  mock-agent [flags]

Flags:
      --command-check-interval int   Command check interval in seconds (default 2)
      --config string                config file
  -h, --help                         help for mock-agent
  -l, --log-level string             Log level (trace, debug, info, warn, error) (default "info")
  -n, --name string                  Agent name (random if not specified)
  -r, --randomized                   Enable randomized behavior mode
  -s, --server string                Fleet backend server URL (default "http://localhost:8080")
      --tick-interval int            Agent tick interval in seconds (default 5)
  -t, --token string                 Authentication token (default "fleet-agent-token-123")
  -w, --worktree string              Agent worktree path (default "/tmp/mock-project")

Usage Examples

Basic Agent

# Simple agent that registers and stays mostly idle
./mock-agent --name "Simple Helper"

Randomized Agent

# Agent with full randomized behavior
./mock-agent --randomized --name "Chaos Tester" --tick-interval 2

Development Testing

# Fast-paced agent for development testing
./mock-agent \
  --randomized \
  --tick-interval 1 \
  --command-check-interval 1 \
  --log-level debug \
  --name "Dev Test Agent"

Production Simulation

# Realistic agent for production-like testing
./mock-agent \
  --randomized \
  --tick-interval 10 \
  --name "Production Simulator" \
  --worktree /var/projects/main-app

Agent Behavior

State Machine

• Idle: Looking for work, may pick up available tasks
• Active: Working on tasks, making progress, occasionally asking questions
• Waiting Feedback: Posted a question and waiting for human response
• Error: Encountered an error, attempting recovery
• Shutting Down: Graceful shutdown in progress

Work Simulation

1. Task Selection: Picks up pending tasks or starts random scenarios
2. Todo Creation: Generates realistic todo lists based on work type
3. Progress Tracking: Simulates incremental progress with metrics
4. Commit Simulation: Periodic commits with realistic messages
5. Question Posting: Context-aware questions when encountering decisions
6. Completion: Marks todos complete and logs success events

Command Processing

• Instructions: stop, continue, restart, focus on security
• Feedback: Positive reinforcement, change requests, error corrections
• Questions: Status inquiries, progress checks, problem diagnosis

Randomized Events

• State transitions based on probability
• Random error scenarios with recovery
• Spontaneous questions and progress updates
• Metric fluctuations and commit activity
• Scenario switching and task completion

Scenarios

Bug Fix - Authentication Issue

• Duration: ~45 minutes
• Steps: Reproduce → Analyze → Fix → Test → Deploy
• Error Rate: 20% (dependency issues, complex JWT logic)
• Question Rate: 15% (session invalidation, user notification)

Feature Development - User Dashboard

• Duration: ~2 hours
• Steps: Requirements → Backend → Frontend → Testing → Documentation
• Error Rate: 10% (framework compatibility, performance)
• Question Rate: 25% (customization level, real-time updates)

Performance Optimization - Database Queries

• Duration: ~90 minutes
• Steps: Identify → Analyze → Optimize → Benchmark → Monitor
• Error Rate: 15% (optimization side effects, memory usage)
• Question Rate: 20% (speed vs consistency, caching strategy)

Security Audit - API Endpoint Review

• Duration: ~3 hours
• Steps: Catalog → Validate → Analyze → Test → Fix → Document
• Error Rate: 30% (false positives, compatibility breaks)
• Question Rate: 35% (breaking changes, compatibility trade-offs)

API Integration

The mock agent fully integrates with the Agent Fleet backend API:

Registration

POST /v1/agents
{
  "name": "Clever Coder",
  "worktree": "/tmp/mock-project"
}

Status Updates

PATCH /v1/agents/{id}
{
  "status": "active",
  "current_task": "Implementing user authentication",
  "progress": 45,
  "files_changed": 8,
  "lines_added": 234,
  "lines_removed": 67
}

Event Logging

POST /v1/agents/{id}/events
{
  "type": "commit",
  "message": "Add user session management",
  "metadata": {
    "files_changed": 3,
    "lines_added": 89
  }
}

Todo Management

POST /v1/agents/{id}/todos
{
  "text": "Implement password validation logic",
  "order": 3
}

Command Processing

GET /v1/agents/{id}/commands?status=sent
PATCH /v1/agents/{id}/commands/{cmd_id}
{
  "status": "completed",
  "response": "Acknowledged. Focusing on security aspects."
}

Development

Adding New Scenarios

1. Edit scenarios/scenarios.go
2. Add new scenario to predefinedScenarios slice
3. Define steps, error conditions, and questions
4. Set appropriate duration and probabilities

Customizing Agent Behavior

• Modify state transition logic in agent/randomized.go
• Adjust probabilities in handleRandomizedBehavior()
• Add new command types in agent/commands.go
• Extend metrics tracking in agent/status.go

Testing Integration

• Start backend server: cd ../backend && ./agent-fleet-backend
• Run multiple agents: ./mock-agent --randomized --name "Agent-$RANDOM"
• Monitor via web UI: http://localhost:8080
• Watch real-time updates via SSE: curl -H "Authorization: Bearer fleet-agent-token-123" http://localhost:8080/v1/stream

Monitoring

The agent logs comprehensive information about its activities:

2025-05-28T15:30:45Z INF Agent registered successfully agent=abc-123 name="Clever Coder"
2025-05-28T15:30:50Z INF Started scenario: Bug Fix - Authentication Issue agent=abc-123 scenario="Bug Fix - Authentication Issue"
2025-05-28T15:31:15Z INF Agent asking question agent=abc-123 question="Should we invalidate all existing user sessions as part of this fix?"
2025-05-28T15:31:20Z INF Processing command agent=abc-123 command=cmd-456 type=feedback content="Yes, invalidate sessions for security"
2025-05-28T15:32:45Z INF Making commit agent=abc-123 message="Fix authentication session validation"
2025-05-28T15:35:30Z INF Completing scenario agent=abc-123 scenario="Bug Fix - Authentication Issue"

This mock agent provides comprehensive simulation of realistic AI coding agent behavior, perfect for testing, demonstration, and development of agent fleet management systems.