Rule reference · Clarity
Hedged instructions
Fires when: "try to", "if possible", "maybe".
Hedges signal that a requirement is negotiable. Models often comply, but under length pressure or with a long context, hedged instructions are the first to be dropped.
Decide whether each item is required. If it is, state it plainly. If it is genuinely optional, say what to do when it cannot be met ("If there is no statistic, write none").
Example that triggers it
Summarize the article below in 3 bullet points. Try to include the author's main argument and, if possible, one supporting statistic.
<article>
{{article}}
</article>info Hedged instructions: "try to", "if possible".
Why it matters. Hedging turns a requirement into an option. The model may treat it as skippable.
Fix. State the requirement directly, or say explicitly that it is optional.
Quality 97/100 · 37 tokens · 1 findings
Fixed version
Summarize the article below in 3 bullet points. Include the author's main argument and one supporting statistic. If the article has no statistic, write "no statistic given".
<article>
{{article}}
</article>Quality 100/100 (+3) · 44 tokens · 0 findings
Resolved by the fix
hedged_instruction: Hedged instructions: "try to", "if possible".
Scores, token counts (GPT-4.1 tokenizer), and findings on this page are computed by the engine when the site is built.
How it affects the score
Each finding subtracts a fixed penalty from 100: critical 30, warning 10, info 3. This rule counts against the Clarity dimension. See the scoring model for the full formula.
Other clarity rules
- Vague language: "something", "stuff", "etc", "and so on".
- Long sentences: A prose sentence over 40 words.
- Negative-only constraints: 3+ prohibitions and no positive requirement.