Run your first AI agent task — from ticket to deployed fix.

Your First Agent

This guide walks you through creating a ticket, assigning it to an AI agent, watching the agent work, reviewing the output, and deploying the fix. By the end, you will have seen the full loop: ticket to deployed code.

1. Create a Ticket

Via the Dashboard

Go to your project in the Codmir dashboard
Click New Ticket
Fill in the details:
- Title: "Fix password reset email not sending"
- Type: Bug
- Priority: High
Click Create

Via the API

import { CodmirClient } from '@codmir/sdk';

const client = new CodmirClient({
  apiKey: process.env.CODMIR_API_KEY,
});

const ticket = await client.createTicket({
  projectId: 'my-project-id',
  title: 'Fix password reset email not sending',
  description: 'Users report that clicking "Reset Password" shows success but no email arrives. Check the email service integration and SMTP configuration.',
  type: 'bug',
  priority: 'high',
});

console.log(`Ticket created: ${ticket.id}`);

2. Assign to an AI Agent

Via the Dashboard

Open the ticket
Click Assign to Agent
Select the agent type (or let Codmir auto-select)
Click Start

Via the API

const task = await client.createAgentTask({
  projectId: 'my-project-id',
  type: 'bug_fix',
  title: 'Fix password reset email not sending',
  description: 'Investigate why password reset emails are not being sent. Check email service, SMTP config, and error logs.',
  priority: 'high',
});

console.log(`Agent task created: ${task.id}`);
console.log(`Status: ${task.status}`);

3. Watch the Agent Work

Once assigned, the agent follows this workflow:

Investigate -- reads error logs, traces the code path, checks configuration
Plan -- creates a task breakdown with specific files to change
Code -- writes the fix, following project conventions from the Cortex Protocol
Test -- runs the test suite against the changes
PR -- opens a pull request with the fix

Monitor Progress via API

// Check task status
const task = await client.getAgentTask('task-id');
console.log(`Status: ${task.status}`);

// Stream real-time updates
const stream = await client.createFrameworkTaskStream('task-id');
stream.subscribe((event) => {
  console.log(`[${event.type}] ${event.data.message || ''}`);
});

Monitor via the Orchestrator

If you have the Overseer orchestrator running, check what the agent is doing:

import { createOrchestrator } from '@codmir/overseer';

const orchestrator = createOrchestrator({
  overseer: {
    projectId: 'my-project-id',
    organizationId: 'my-org',
    apiKey: process.env.CODMIR_API_KEY!,
  },
  defaultRepoUrl: 'https://github.com/my-org/my-app',
  defaultBranch: 'main',
});

// "What are you working on?"
const summary = orchestrator.getStatusSummary();
console.log(summary);

// Is it already handling this ticket?
const handling = orchestrator.isWorkingOn('ticket-id');
console.log(`Already on it: ${handling}`);

4. Review and Approve

When the agent finishes, you receive a notification. The pull request includes:

A description of the root cause
The code changes with explanations
Test results
A link back to the original ticket

Review the PR as you would any human-authored code:

// List agent tasks and their results
const tasks = await client.listAgentTasks('my-project-id');

for (const task of tasks) {
  if (task.status === 'completed') {
    console.log(`${task.title}: ${task.status}`);
  }
}

5. Auto-Deploy

If your CI/CD pipeline is connected, merging the PR triggers an automatic deployment. Codmir tracks the deployment and monitors for regressions.

If a regression is detected after deployment:

Overseer captures the new errors
The DecisionEngine evaluates severity
If critical, the fix is reverted automatically
You are notified via the configured escalation chain

The Full Loop

Ticket Created
    ↓
Agent Assigned
    ↓
Investigate → Plan → Code → Test → PR
    ↓
Human Review
    ↓
Merge → Deploy → Monitor
    ↓
(Regression?) → Auto-Revert → Escalate

Using Developer Agents

For more control over the agent's behavior, use the DevTeam from @codmir/ai:

import { createDevTeam } from '@codmir/ai';

const team = createDevTeam({
  projectContext: {
    projectId: 'my-project',
    name: 'My App',
    techStack: ['TypeScript', 'Next.js', 'Prisma'],
  },
});

const result = await team.assignTicket({
  id: 'TICKET-42',
  title: 'Fix password reset email',
  description: 'Password reset emails are not sending',
  priority: 'high',
  status: 'open',
});

console.log(result.plan);    // Task breakdown
console.log(result.changes); // Files modified
console.log(result.tests);   // Tests written or updated

Next Steps

Self-Healing -- let agents fix bugs automatically without creating tickets
Swarm Execution -- scale agent work across thousands of workers
Cortex Protocol -- give agents deep project context for better fixes

Run your first AI agent task — from ticket to deployed fix.

Your First Agent

1. Create a Ticket

Via the Dashboard

Go to your project in the Codmir dashboard
Click New Ticket
Fill in the details:
- Title: "Fix password reset email not sending"
- Type: Bug
- Priority: High
Click Create

Via the API

import { CodmirClient } from '@codmir/sdk';

const client = new CodmirClient({
  apiKey: process.env.CODMIR_API_KEY,
});

const ticket = await client.createTicket({
  projectId: 'my-project-id',
  title: 'Fix password reset email not sending',
  description: 'Users report that clicking "Reset Password" shows success but no email arrives. Check the email service integration and SMTP configuration.',
  type: 'bug',
  priority: 'high',
});

console.log(`Ticket created: ${ticket.id}`);

2. Assign to an AI Agent

Via the Dashboard

Open the ticket
Click Assign to Agent
Select the agent type (or let Codmir auto-select)
Click Start

Via the API

const task = await client.createAgentTask({
  projectId: 'my-project-id',
  type: 'bug_fix',
  title: 'Fix password reset email not sending',
  description: 'Investigate why password reset emails are not being sent. Check email service, SMTP config, and error logs.',
  priority: 'high',
});

console.log(`Agent task created: ${task.id}`);
console.log(`Status: ${task.status}`);

3. Watch the Agent Work

Once assigned, the agent follows this workflow:

Investigate -- reads error logs, traces the code path, checks configuration
Plan -- creates a task breakdown with specific files to change
Code -- writes the fix, following project conventions from the Cortex Protocol
Test -- runs the test suite against the changes
PR -- opens a pull request with the fix

Monitor Progress via API

// Check task status
const task = await client.getAgentTask('task-id');
console.log(`Status: ${task.status}`);

// Stream real-time updates
const stream = await client.createFrameworkTaskStream('task-id');
stream.subscribe((event) => {
  console.log(`[${event.type}] ${event.data.message || ''}`);
});

Monitor via the Orchestrator

If you have the Overseer orchestrator running, check what the agent is doing:

import { createOrchestrator } from '@codmir/overseer';

const orchestrator = createOrchestrator({
  overseer: {
    projectId: 'my-project-id',
    organizationId: 'my-org',
    apiKey: process.env.CODMIR_API_KEY!,
  },
  defaultRepoUrl: 'https://github.com/my-org/my-app',
  defaultBranch: 'main',
});

// "What are you working on?"
const summary = orchestrator.getStatusSummary();
console.log(summary);

// Is it already handling this ticket?
const handling = orchestrator.isWorkingOn('ticket-id');
console.log(`Already on it: ${handling}`);

4. Review and Approve

When the agent finishes, you receive a notification. The pull request includes:

A description of the root cause
The code changes with explanations
Test results
A link back to the original ticket

Review the PR as you would any human-authored code:

// List agent tasks and their results
const tasks = await client.listAgentTasks('my-project-id');

for (const task of tasks) {
  if (task.status === 'completed') {
    console.log(`${task.title}: ${task.status}`);
  }
}

5. Auto-Deploy

If your CI/CD pipeline is connected, merging the PR triggers an automatic deployment. Codmir tracks the deployment and monitors for regressions.

If a regression is detected after deployment:

Overseer captures the new errors
The DecisionEngine evaluates severity
If critical, the fix is reverted automatically
You are notified via the configured escalation chain

The Full Loop

Ticket Created
    ↓
Agent Assigned
    ↓
Investigate → Plan → Code → Test → PR
    ↓
Human Review
    ↓
Merge → Deploy → Monitor
    ↓
(Regression?) → Auto-Revert → Escalate

Using Developer Agents

For more control over the agent's behavior, use the DevTeam from @codmir/ai:

import { createDevTeam } from '@codmir/ai';

const team = createDevTeam({
  projectContext: {
    projectId: 'my-project',
    name: 'My App',
    techStack: ['TypeScript', 'Next.js', 'Prisma'],
  },
});

const result = await team.assignTicket({
  id: 'TICKET-42',
  title: 'Fix password reset email',
  description: 'Password reset emails are not sending',
  priority: 'high',
  status: 'open',
});

console.log(result.plan);    // Task breakdown
console.log(result.changes); // Files modified
console.log(result.tests);   // Tests written or updated

Next Steps

Self-Healing -- let agents fix bugs automatically without creating tickets
Swarm Execution -- scale agent work across thousands of workers
Cortex Protocol -- give agents deep project context for better fixes