Clarity, Specificity & Positive Framing

Be Clear, Direct, and Specific

State the task, constraints, and desired output explicitly. Ambiguity is the top cause of poor output -- a vague or terse instruction leaves the model to guess at intent, and that guess varies from run to run. Tell Claude what to do, not just what to avoid.

Positive Framing Over Negative Lists

A positive instruction -- "respond only in JSON matching this schema" -- steers more reliably than a long list of "don'ts" (don't add commentary, don't use markdown, don't include extra fields, ...). Positive framing tells the model what the target looks like; negative lists only narrow what's excluded, leaving the actual target under-specified. "More prohibitions = safer" is a tempting but wrong instinct.

Output Constraints

Specify the exact shape you need: "respond only with JSON matching this schema," "no preamble." Constrain length and structure explicitly rather than leaving format to inference -- the more precisely the desired output is described, the less variance the model has room to introduce.

Common exam traps

• "More prohibitions = safer." Positive, specific instructions steer better than long "don't" lists.
• Assuming a vague or terse instruction is fine because "the model will figure it out" -- ambiguity is the top cause of poor output, not a minor stylistic choice.