Teach the model by example: supply 2-3 input to output samples, then have it apply the pattern to your task.
Prompts / Techniques / Prompt Compression Optimizer for High-Volume API Calls
Prompt Compression Optimizer for High-Volume API Calls
Rewrites a verbose production prompt into a token-lean version while preserving behavior and guardrails.
ROLE: You are a prompt-optimization specialist focused on cutting token cost without behavior drift in production LLM pipelines.
CONTEXT: Here is a current prompt run [CALLS_PER_DAY] times daily: [ORIGINAL_PROMPT]. Its required behaviors are [MUST_KEEP_BEHAVIORS] and its hard guardrails are [GUARDRAILS].
TASK:
1. Identify redundant instructions, restatable defaults, and removable politeness or filler.
2. Produce a compressed version that preserves every must-keep behavior and every guardrail.
3. Build a before/after table noting what was cut and why it is safe to cut.
4. List 3 behaviors at risk of regressing and the test that would catch each.
CONSTRAINTS: Do not weaken any guardrail to save tokens. Keep all [BRACKETED] placeholders intact. State the approximate token reduction. If a behavior cannot be safely compressed, keep it and say so.
OUTPUT FORMAT: COMPRESSED PROMPT code block, CHANGE TABLE (Removed, Reason, Risk), REGRESSION TESTS list, estimated token delta.