REST API Reference

For languages without a first-class SDK, or for bare-metal implementations, you can interface directly with the /optimize REST endpoint.

The endpoint expects standard application/json data and Bearer token authentication.

HTTP Request


POST /optimize HTTP/1.1
Host: api.trunkate.ai
Authorization: Bearer tk_live_...
Content-Type: application/json

Request Payload

JSON Key	Type	Required	Description
`text`	string	Yes	The natural language prompt to optimize.
`task`	string	No	Optional guidance for preservation ranking.
`budget`	`number\|string`	No	Absolute tokens (e.g. `500`) or percentage (e.g. `"20%"`).
`model`	string	No	Target LLM (default: `gpt-4o`).

Response Body


{
  "optimized_text": "The optimized prompt here...",
  "original_tokens": 120,
  "optimized_tokens": 45,
  "reduction_percent": 62,
  "latency_ms": 12.5,
  "soft_limit_enforced": false
}

JSON Key	Type	Description
`optimized_text`	string	The resulting optimized prompt.
`original_tokens`	number	Input token count.
`optimized_tokens`	number	Output token count.
`reduction_percent`	number	Percentage of tokens saved.
`latency_ms`	number	Processing time in milliseconds.
`soft_limit_enforced`	boolean	`true` if the requested budget was too small and a fallback was used.

Example using cURL


# OPTION A: Fixed Token Budget
curl -X POST https://api.trunkate.ai/optimize \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text": "...", "budget": 500}'
 
# OPTION B: Proactive Percentage Buffer
curl -X POST https://api.trunkate.ai/optimize \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text": "...", "budget": "20%"}'