gepa_record_trajectory
Track and log detailed execution steps, results, and metadata for prompt evaluations to analyze performance and optimize AI prompts using data-driven insights.
Instructions
Record execution trajectory for prompt evaluation
Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| executionSteps | Yes | Sequence of execution steps | |
| metadata | No | Additional execution metadata (optional) | |
| promptId | Yes | Unique identifier for the prompt candidate | |
| result | Yes | Final execution result and performance score | |
| taskId | Yes | Identifier for the specific task instance |
Input Schema (JSON Schema)
{
"properties": {
"executionSteps": {
"description": "Sequence of execution steps",
"items": {
"properties": {
"action": {
"type": "string"
},
"input": {
"type": "object"
},
"output": {
"type": "object"
},
"success": {
"type": "boolean"
},
"timestamp": {
"type": "string"
}
},
"required": [
"action",
"timestamp",
"success"
],
"type": "object"
},
"type": "array"
},
"metadata": {
"description": "Additional execution metadata (optional)",
"properties": {
"executionTime": {
"type": "number"
},
"llmModel": {
"type": "string"
},
"tokenUsage": {
"type": "number"
}
},
"type": "object"
},
"promptId": {
"description": "Unique identifier for the prompt candidate",
"type": "string"
},
"result": {
"description": "Final execution result and performance score",
"properties": {
"output": {
"type": "object"
},
"score": {
"type": "number"
},
"success": {
"type": "boolean"
}
},
"required": [
"success",
"score"
],
"type": "object"
},
"taskId": {
"description": "Identifier for the specific task instance",
"type": "string"
}
},
"required": [
"promptId",
"taskId",
"executionSteps",
"result"
],
"type": "object"
}