output-length-missing — Output Length Constraint

Severity: INFO · Auto-fix: No · Category: 🏗️ Structure

What It Does

Detects prompts that contain a task verb (write, generate, list, etc.) but don't specify how long the output should be. Without a length constraint, response length varies unpredictably between model versions, calls, and providers — a consistent response length requires an explicit constraint.

Trigger Conditions

Three conditions must all be true

1. The prompt is at least 80 characters long
2. The prompt contains a task verb from the list below
3. The prompt does not contain any length constraint from the list below

Task Verbs (trigger the check)

write, create, implement, build, generate, list, extract, return, output, summarize, describe, explain, produce, draft, compose

Length Constraint Keywords (suppress the check)

Pattern	Examples
Explicit counts	200 words, 3 sentences, 5 bullet points, 50 tokens
Maximum/limit	max 100 words, maximum length, word limit, token limit
Brevity directives	be brief, be concise, keep it short, succinct, terse
Relative bounds	no more than 300 words, at most 3 paragraphs, up to 5 items
Other	truncate, limited to 200, in 3 points

Examples

Triggers the rule

Write a detailed explanation of how transformers work in neural networks.
Include the attention mechanism, positional encoding, and layer normalization.

Has write — no length constraint. Output could be 100 or 2,000 words.

[ INFO ] output-length-missing (line -)
  No output length constraint specified. Add a word/token/sentence limit
  for consistent response lengths.

Passes the rule

Write a detailed explanation of how transformers work in neural networks
in no more than 300 words. Cover the attention mechanism, positional
encoding, and layer normalization.

Has no more than 300 words → no finding.

Create a 5-bullet summary of the document below.

Has 5-bullet → no finding.

Generate a concise overview of the project status.

Has concise → no finding.

False Positives

Short prompts — prompts under 80 characters don't fire this rule. Write a haiku about rain (24 chars) won't trigger.

Prompts with implicit constraints — "Write a tweet announcing the launch" implicitly requires brevity (Twitter's 280-char limit), but promptlint doesn't understand domain conventions. Add "in 240 characters or fewer" to be explicit.

Template variables — if your length constraint is injected via a variable ({length_limit}), promptlint won't see it. Use # promptlint-disable output-length-missing to suppress.

Configuration

rules:
  output_length_missing: true   # or false to disable

Promote to WARN:

rules:
  output_length_missing:
    enabled: true
    severity: warn

Disable for a single prompt:

Write an essay about climate change.
# promptlint-disable output-length-missing

Recommended Fix

Add an explicit length boundary in one of these forms:

# Word count
Write a 200-word summary of the following document.

# Sentence count
Explain the concept in 3 sentences.

# Bullet points
List the key advantages in 5 bullet points.

# Token budget (for API cost control)
Summarize the article in at most 100 tokens.

# Brevity directive
Describe the architecture. Be concise.

For production prompts, word counts or token limits are preferable over brevity directives — they're more precise and work consistently across model versions.

Why Length Constraints Matter

Without a length constraint:

• A model update can silently change average response length by 30–50%
• Responses in production vary from a single sentence to multiple paragraphs for the same prompt
• Long responses increase latency and token cost unexpectedly
• Short responses may miss required content
• Downstream parsing that expects a fixed structure becomes brittle

Adding a length constraint doesn't prevent good answers — it creates a contract between your prompt and the model, making outputs auditable and reproducible.