{"ID":3052317,"CreatedAt":"2026-06-04T04:41:36.695875263Z","UpdatedAt":"2026-06-06T05:11:44.966731765Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2606.04465","arxiv_id":"2606.04465","title":"SePO: Self-Evolving Prompt Agent for System Prompt Optimization","abstract":"System prompt optimization improves agent behavior without modifying the underlying model, yielding human-readable, model-agnostic instructions. Existing methods build a prompt agent that refines task agents' system prompts, yet leave the prompt agent's own system prompt hand-engineered and fixed. We propose Self-Evolving Prompt Optimization (SePO), which treats the prompt agent's own system prompt as an optimization target alongside task agents' system prompts. SePO adopts a self-referential design. A single prompt agent improves both task agents' system prompts and its own under an open-ended evolutionary search that maintains an archive of candidate prompts as stepping stones. Training proceeds in two stages: pre-training evolves the prompt agent on a multi-task pool, and fine-tuning then applies it to a target task. Across five benchmarks spanning math (AIME'25), abstract reasoning (ARC-AGI-1), graduate-level science (GPQA), code generation (MBPP), and logic puzzles (Sudoku), SePO consistently outperforms Manual-CoT, TextGrad, and MetaSPO, improving the average accuracy by 4.49 points compared to Manual-CoT. The prompt optimization skill from pre-training also generalizes to tasks beyond the pre-training mixture, rather than memorizing per-task prompts.","short_abstract":"System prompt optimization improves agent behavior without modifying the underlying model, yielding human-readable, model-agnostic instructions. Existing methods build a prompt agent that refines task agents' system prompts, yet leave the prompt agent's own system prompt hand-engineered and fixed. We propose Self-Evolv...","url_abs":"https://arxiv.org/abs/2606.04465","url_pdf":"https://arxiv.org/pdf/2606.04465v1","authors":"[\"Wangcheng Tao\",\"Han Wu\",\"Weng-Fai Wong\"]","published":"2026-06-03T05:20:30Z","proceeding":"cs.CL","tasks":"[\"cs.CL\",\"cs.AI\"]","methods":"[]","has_code":false}