PromptLab Public API Free

AI-powered prompt improvement and grading endpoints — no API key required, no signup.

Base URL: http://143.198.136.81:8802 | Rate limit: 20 requests / hour per IP | Auth: None required | Format: JSON

Endpoints

POST /tools/improve

Improve any weak or vague prompt using AI. Returns the rewritten prompt alongside token usage and cost. Powered by Claude Haiku.

Request body

Field	Type	Description
prompt	string	The prompt to improve. Max 500 characters.

Response fields

Field	Type	Description
original	string	The prompt you submitted.
improved	string	The AI-rewritten, more specific prompt.
tokens_used	integer	Total input + output tokens consumed.
cost_usd	float	Actual cost in USD for this call.
share_url	string	Shareable permalink to this result.

curl example

curl -s -X POST http://143.198.136.81:8802/tools/improve   -H "Content-Type: application/json"   -d '{"prompt": "write a blog post about AI"}'

Example response

{
  "original": "write a blog post about AI",
  "improved": "Write a 600-word blog post for a technical audience explaining how large language models generate text. Include one concrete analogy, three practical use cases, and a short conclusion with a call to action.",
  "tokens_used": 87,
  "cost_usd": 0.00032,
  "share_url": "http://143.198.136.81:8802/share/abc123"
}

POST /tools/grade

Grades a prompt A–F across five quality dimensions: context, specificity, constraints, examples, and goal clarity. Returns a total score out of 10, a letter grade, the top improvement action, and a fully rewritten improved prompt.

Request body

Field	Type	Description
prompt	string	The prompt to grade. Max 1000 characters.

Response fields

Field	Type	Description
original	string	The prompt you submitted.
total_score	integer	Score from 0–10.
max_score	integer	Always 10.
grade	string	Letter grade: A, B, C, D, or F.
scores	object	Per-dimension scores (0–2 each): `context`, `specificity`, `constraints`, `examples`, `goal_clarity`.
top_improvement	string	The single most impactful change you could make.
improved_prompt	string	A fully rewritten version of your prompt with all gaps filled.
cost_usd	float	Actual cost in USD for this call.
share_url	string	Shareable permalink to this result.

curl example

curl -s -X POST http://143.198.136.81:8802/tools/grade   -H "Content-Type: application/json"   -d '{"prompt": "summarize this document"}'

Example response

{
  "original": "summarize this document",
  "total_score": 3,
  "max_score": 10,
  "grade": "D",
  "scores": {
    "context": 1, "specificity": 0,
    "constraints": 0, "examples": 1, "goal_clarity": 1
  },
  "top_improvement": "Specify the desired output length and format (e.g. bullet list, max 200 words)",
  "improved_prompt": "Summarize the following document in 3–5 bullet points, each under 30 words. Focus on key decisions, outcomes, and action items. Omit background context.",
  "cost_usd": 0.00041,
  "share_url": "http://143.198.136.81:8802/share/def456"
}

POST /promptlab/compare

Runs your prompt through both Claude Haiku and Claude Sonnet in parallel and returns both responses alongside a side-by-side cost comparison. Useful for deciding which model tier is sufficient for your use case. Requires a PromptLab access key (use demo to try it).

Request body

Field	Type	Description
prompt	string	The prompt to run through both models.
api_key	string	PromptLab access key. Use `"demo"` for free access.

Response fields

Field	Type	Description
results.haiku	object	Haiku response, token counts, cost_usd, latency_ms.
results.sonnet	object	Sonnet response, token counts, cost_usd, latency_ms.
cost_comparison	object	haiku_cost, sonnet_cost, savings_with_haiku_usd, savings_with_haiku_pct.

curl example

curl -s -X POST http://143.198.136.81:8802/promptlab/compare   -H "Content-Type: application/json"   -d '{"prompt": "Explain async/await in Python in one paragraph.", "api_key": "demo"}'

Example response

{
  "prompt": "Explain async/await in Python in one paragraph.",
  "results": {
    "haiku":  { "response": "...", "cost_usd": 0.00012, "latency_ms": 480 },
    "sonnet": { "response": "...", "cost_usd": 0.00063, "latency_ms": 820 }
  },
  "cost_comparison": {
    "haiku_cost": 0.00012,
    "sonnet_cost": 0.00063,
    "savings_with_haiku_usd": 0.00051,
    "savings_with_haiku_pct": 81.0
  }
}

Python Integration Example

Drop this into any script to grade prompts programmatically before sending them to a production LLM. It adds about 0.4 ms of overhead and costs under $0.001 per call.

import requests

def grade_prompt(prompt: str) -> dict:
    """
    Grade a prompt A-F using the PromptLab free API.
    Returns grade, total_score/10, and a rewritten improved_prompt.
    """
    url = "http://143.198.136.81:8802/tools/grade"
    resp = requests.post(url, json={"prompt": prompt}, timeout=15)
    resp.raise_for_status()
    return resp.json()


# Example: fail fast if prompt quality is below a threshold
result = grade_prompt("summarize this document")

if result["grade"] in ("D", "F"):
    print(f"Weak prompt (grade {result['grade']}, score {result['total_score']}/10)")
    print(f"Top fix: {result['top_improvement']}")
    print(f"Suggested rewrite:
{result['improved_prompt']}")
else:
    print(f"Prompt grade: {result['grade']} ({result['total_score']}/10) — sending to model.")
    # proceed with your LLM call here

Rate Limits & Fair Use

20 requests per hour per IP across all endpoints combined.
No authentication is required for /tools/improve and /tools/grade.
/promptlab/compare requires a PromptLab access key — use demo to get started.
Input length limits: 500 characters for /tools/improve, 1000 characters for /tools/grade.
This API is provided free for experimentation and integration. Please do not use it for bulk scraping or high-frequency automated pipelines without reaching out first.

Need 130+ production-ready prompts?

PromptLab packs cover product management, software development, and startup strategy — battle-tested and ready to drop into your workflow.

Browse prompt packs at PromptLab