AgentSkillsCN

prompt_evolution

通过并行子代理、LLM作为“裁判”的评分机制，以及迭代式的筛选优化，不断进化提示词。当您需要针对特定任务优化提示词，并拥有清晰的评估标准时，此方法将助您事半功倍。但请勿将其用于一次性提示词，或当评估标准尚不明确时。

SKILL.md
--- frontmatter
name: prompt_evolution
description: >
  Evolve prompts through genetic-style optimization using parallel sub-agents,
  LLM-as-judge scoring, and iterative selection. Use when you need to optimize
  a prompt for a specific task and have a clear evaluation criterion.
  Do NOT use for one-off prompts or when the criterion is unclear.
allowed-tools: Bash(uv run *)

Optimize prompts through evolutionary search: generate variants, score with llm_judge, keep the fittest, mutate and recombine, repeat.

The Loop

1. Setup

Define before starting:

  • •Seed prompt: the starting prompt to optimize
  • •Test inputs: 3-5 representative inputs the prompt must handle
  • •Judge criterion: a specific, binary pass/fail criterion (see llm_judge skill)
  • •Generations: how many rounds to run (default: 3)
  • •Population size: variants per generation (default: 4)

2. Generate variants

Launch parallel sub-agents, each using discussion_partners to produce one variant. Apply two genetic operators:

Mutation (always): Literally change words in the prompt. Swap synonyms, delete filler, add specificity, reorder sentences, strengthen or weaken qualifiers. Each mutation should change 1-5 words or phrases — small, targeted edits, not wholesale rewrites.

Here is a prompt: [SEED]. Mutate it: change 1-5 specific words or phrases to potentially improve [CRITERION]. Keep the core intent. Return ONLY the new prompt text.

Crossover (generation 2+): Literally splice sections from two parent prompts. Split each parent into numbered sections (by paragraph or logical block), then take section 1 from parent A, section 2 from parent B, etc. — producing a child that inherits concrete text from both parents.

Here are two prompts that scored well: Parent A: [PROMPT_A] Parent B: [PROMPT_B] Split each into sections. Combine: take some sections from A and some from B to create a child prompt. Then apply 1-3 word-level mutations. Return ONLY the new prompt text.

For generation 1, all variants mutate from the seed. For later generations, half the population uses crossover from the top 2 parents, half uses mutation from a random survivor.

3. Score each variant

For each variant, for each test input:

  1. •Run the variant prompt against a test model (via discussion_partners)
  2. •Score the output using llm_judge with the defined criterion
  3. •Record pass/fail per input

Variant fitness = pass rate across all test inputs.

4. Select

Keep the top 50% of variants (minimum 2). These become parents for the next generation.

5. Repeat

Mutate and recombine the surviving variants to fill the population. Run for the specified number of generations.

6. Return the winner

Output the highest-fitness prompt with its score and the specific test results.

Example

code
Seed: "Summarize this article in 3 bullet points"
Criterion: "Each bullet captures a distinct key point, no redundancy, no hallucinated claims"
Test inputs: [3 different articles]
Generations: 3, Population: 4

Generation 1: 4 variants of the seed, scored → keep top 2 Generation 2: 2 survivors + 2 offspring (mutated/recombined), scored → keep top 2 Generation 3: same → return winner

Sub-Agent Orchestration

Each scoring round involves population × test_inputs parallel sub-agent calls. Structure:

  1. •Variant generation: parallel sub-agents, each uses discussion_partners to create one variant
  2. •Output generation: parallel sub-agents, each runs one variant × one test input via discussion_partners
  3. •Judging: parallel sub-agents, each scores one output via llm_judge pattern

Maximize parallelism at each step. Wait for all results before proceeding to selection.

When to Stop Early

  • •If the top variant achieves 100% pass rate for 2 consecutive generations
  • •If fitness plateaus (no improvement for 2 generations)