Generate Template

Generate an institutional document template for Mexican government entities

POST /api/generate-template

Generate Template

cURL

curl --request POST \
  --url https://api.example.com/api/generate-template \
  --header 'Content-Type: application/json' \
  --data '
{
  "prompt": "<string>",
  "area": "<string>"
}
'

200

{
  "text": "<string>"
}

Endpoint

POST /api/generate-template

Generates a formatted institutional document template (machote) based on the provided prompt and area. The template includes standard elements like institutional header, subject line, date/location, body, signature block, and cc field.

Request Body

prompt · string · required

Description of the document to generate. This should specify the purpose and content of the institutional document.

Example: "Oficio de solicitud de material de oficina para el departamento"

area · string

The governmental area or department requesting the document. If not provided, defaults to “No especificada”.

Example: "Recursos Humanos", "Obras Públicas", "Secretaría General"

Response

text · string

The generated institutional document template. Contains the complete formatted document with variable placeholders in {{Variable}} format (double braces).

The template includes:

• Institutional header
• ASUNTO (subject line)
• Place and date
• Document body
• Closing
• ATENTAMENTE
• Signature block
• C.c.p. (copy to)

System Prompt

The API uses a specialized system prompt that instructs the LLM to:

• Act as an institutional writing assistant for a Mexican City Hall (Ayuntamiento)
• Return ONLY the final template text (no explanations)
• Include all standard institutional document elements
• Use variables in {{Variable}} format with double braces (not brackets or single braces)

This is implemented in server.js:9-16 via the buildSystem() function.

Examples

cURL

curl -X POST http://localhost:5055/api/generate-template \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Oficio de solicitud de material de oficina para el departamento",
    "area": "Recursos Humanos"
  }'

JavaScript (fetch)

const response = await fetch('http://localhost:5055/api/generate-template', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({
    prompt: 'Oficio de solicitud de material de oficina para el departamento',
    area: 'Recursos Humanos'
  })
});

const data = await response.json();
console.log(data.text);

Success Response

{
  "text": "AYUNTAMIENTO DE {{Municipio}}\n\nASUNTO: Solicitud de Material de Oficina\n\n{{Lugar}}, {{Fecha}}\n\n{{Destinatario}}\n{{Cargo}}\nPresente\n\nPor medio del presente, el Departamento de Recursos Humanos solicita...\n\nATENTAMENTE\n\n{{Firma}}\n{{Cargo}}\n\nC.c.p. {{Copia}}"
}

Error Responses

Show 400 Bad Request

Returned when the prompt parameter is missing or not a string.

{
  "error": "prompt es requerido"
}

Cause: Missing or invalid prompt in request body (server.js:21-23)

Show 502 Bad Gateway

Returned when the Groq API request fails.

{
  "error": "Groq error",
  "details": "...error details from Groq API..."
}

Cause: Groq API returned a non-OK response (server.js:43-46)

Common reasons:

• Invalid or missing GROQ_API_KEY environment variable
• Groq API rate limits exceeded
• Groq API service issues

Show 500 Internal Server Error

Returned when an unexpected server error occurs.

{
  "error": "server error",
  "details": "...exception message..."
}

Cause: Unhandled exception during request processing (server.js:52)

Implementation Details

The endpoint (server.js:18-54):

1. Validates the prompt parameter is a string
2. Constructs messages array with system prompt and user request
3. Calls Groq’s API with model llama-3.1-8b-instant and temperature 0.4
4. Extracts the generated text from data.choices[0].message.content
5. Returns the trimmed text in the response

Model Configuration

• Model: llama-3.1-8b-instant
• Temperature: 0.4 (balanced between consistency and creativity)
• API: Groq OpenAI-compatible endpoint