agentPlasticity Documentation

Track and evaluate your AI agent calls with a single API call. Works with any language or framework.

Quick Start

Send a POST request to /api/ingest with your agent API token. That's it — works with any language or framework.

Create a project and agent in Projects
Copy the agent's API token
Send your AI call data using any HTTP client

JavaScript / TypeScript

track.ts

await fetch('https://www.agentplasticity.com/api/ingest', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    'Authorization': 'Bearer YOUR_API_TOKEN',
  },
  body: JSON.stringify({
    prompt: 'Summarize this document...',
    output: 'The document discusses...',
    modelName: 'gpt-4o',
  }),
});

With optional metadata (token usage, latency):

track-with-metadata.ts

const start = Date.now();
const output = await llm.chat('Summarize this document...');
const latencyMs = Date.now() - start;

await fetch('https://www.agentplasticity.com/api/ingest', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    'Authorization': `Bearer ${process.env.AGENT_PLASTICITY_API_KEY}`,
  },
  body: JSON.stringify({
    prompt: 'Summarize this document...',
    output,
    modelName: 'gpt-4o',
    systemPrompt: 'You are a helpful assistant.',
    metadata: { latencyMs, tokenUsage: 350 },
  }),
});

Python

track.py

import requests

requests.post(
    "https://www.agentplasticity.com/api/ingest",
    headers={"Authorization": "Bearer YOUR_API_TOKEN"},
    json={
        "prompt": "Summarize this document...",
        "output": "The document discusses...",
        "modelName": "gpt-4o",
        "metadata": {"latencyMs": 1200, "tokenUsage": 350},
    },
)

cURL

Terminal

curl -X POST https://www.agentplasticity.com/api/ingest \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -d '{
    "prompt": "Summarize this document...",
    "output": "The document discusses...",
    "modelName": "gpt-4o",
    "metadata": {
      "latencyMs": 1200,
      "tokenUsage": 350
    }
  }'

Request Body

All requests go to POST /api/ingest with a Bearer token in the Authorization header.

Field	Type	Required	Description
prompt	string	Yes	The prompt sent to the AI model
output	string	Yes	The AI model's response
modelName	string	Yes	Model identifier (e.g. "gpt-4o")
systemPrompt	string	No	System prompt used for the call
metadata	object	No	Optional data: tokenUsage (number), latencyMs (number), or any custom fields

Error Codes

The /api/ingest endpoint returns standard HTTP status codes:

Status	Meaning	What to do
201	Created	Call tracked successfully
400	Bad Request	Check required fields: prompt, output, modelName
401	Unauthorized	Check your API token is valid and the agent is active
429	Rate Limited	Back off and retry. Check X-RateLimit-Reset header
500	Server Error	Retry with exponential backoff

agentPlasticityGo to Projects