Managing Agents

Organize, update, and maintain your agents over time.

Viewing Agents

Dashboard

Visit converra.ai/agents to see all your agents with:

• Status (active, draft, archived)
• Last updated date
• Performance metrics
• Tags

Via MCP

Show me my agents

List agents tagged with "production"

Via SDK

const { data: agents } = await converra.agents.list();

agents.forEach(a => {
  console.log(`${a.name} (${a.llmModel}) - ${a.status}`);
});

Updating Agents

Dashboard

1. Click on an agent to open it
2. Edit the system prompt
3. Click Save

Previous versions are automatically saved.

Via MCP

Update my support agent to be more friendly and add a greeting

Via SDK

await converra.agents.update('agent_123', {
  content: 'Updated system prompt...',
  description: 'Now with improved greeting'
});

Organizing with Tags

Tags help you filter and organize agents:

await converra.agents.update('agent_123', {
  tags: ['production', 'support', 'v2']
});

Common tag patterns:

• Environment: production, staging, development
• Team: support, sales, marketing
• Version: v1, v2, experimental
• Status: active, deprecated, testing

Agent Status

Status	Meaning
draft	In development, not used in production
active	Currently in use
deprecated	Scheduled for removal
archived	No longer in use, preserved for reference

Version History

Converra automatically tracks agent versions:

• Every save creates a new version
• Applied optimization variants create versions
• You can view and compare versions in the dashboard

Duplicating Agents

To create a copy of an existing agent:

const original = await converra.agents.get('agent_123');

const copy = await converra.agents.create({
  name: `${original.name} (Copy)`,
  content: original.content,
  llmModel: original.llmModel,
  tags: [...original.tags, 'copy']
});

Deleting Agents

WARNING

Deleting an agent is permanent and will affect any applications using it.

await converra.agents.delete('agent_123');

Consider archiving instead:

await converra.agents.update('agent_123', {
  status: 'archived'
});

Best Practices

1. Use descriptive names - Make it clear what each agent does
2. Tag consistently - Establish a tagging convention with your team
3. Document changes - Use the description field to note why changes were made
4. Archive, don't delete - Keep history for reference
5. Review regularly - Check agent performance monthly

Next Steps

• Best Practices - Write better system prompts
• Running Optimizations - Improve agents
• Logging Runs - Track performance