Prompt manager

Prompt Manager is where you create and organize custom prompts, configure LLM providers and models, and manage AI execution settings. It’s organized into four tabs: Categories, Models, Templates, and Execution.

In this guide:

• Create and organize prompts by category
• Configure LLM models and providers
• Save and use prompt templates
• Set execution preferences

Step 1: Access Prompt Manager

Visit the direct URL /prompts-manager to open Prompt Manager.

Categories tab

Organize your prompts by topic. Create categories like “Inventory Analysis”, “Ad Performance”, or “Customer Insights” and assign prompts to them.

Models tab

Configure your LLM provider connections (OpenAI, OpenRouter).

• Add a provider — connect your API keys for OpenAI or OpenRouter.
• Test the connection — verify your API key works with the provider.

For details on setting up LLM connections, see Multi-LLM connections.

Templates tab

View and manage prompt templates. Templates let you save a prompt structure with placeholders (e.g., {{marketplace}}, {{date_range}}) so your team reuses the same prompt across different contexts.

• Create a template — write a prompt with variables and save it.
• Edit status — templates can be ACTIVE, ARCHIVED, or DELETED.
• Assign models — select which LLM model(s) are compatible with this template.

Execution tab

Test and execute prompts in three modes:

• Template execution — run a saved template with your inputs.
• Raw prompt execution — test an ad-hoc prompt directly.
• Batch execution — run prompts on multiple inputs at once.

Troubleshooting

• Can’t connect my LLM provider. Check that your API key is valid and has the required permissions. Use the Test button in the Models tab to verify.
• Can’t find a template. Make sure the template status is ACTIVE (not ARCHIVED or DELETED).
• Execution fails. Verify the template variables are filled correctly and the selected model is available and connected.

Related articles