FlowEvo: Self-Evolving Agents through the Co-Evolution of Workflows and Executable Skills
Authors: Zeyu Ren, Ling Yue, Ran Li, Yishu Wang, Shengxiang Xu, Hanmo Liu, Shaowu Pan, Shimin Di
Submitted: 18 April 2026
Abstract
Large language model (LLM) agents increasingly solve complex tasks by constructing inference-time workflows that integrate reasoning, tool use, and code execution. While such workflows enable flexible problem solving, the useful procedures discovered during execution are often transient—they help solve the current task but are not retained in a form that can systematically benefit future tasks.
We present FlowEvo, a training-free framework that compiles successful execution traces into reusable skill records. Each record pairs a callable artifact with auxiliary structured guidance, and admission applies interface, replay, and safety checks where feasible. These skill records persist in a skill bank at inference time.
FlowEvo is organized around three coupled mechanisms:
- Workflow-to-Skill Compilation – extracts reusable executable artifacts from successful traces.
- Skill-to-Workflow Feedback – retrieves accumulated skills to support future problem solving through either direct execution or structured context injection.
- Skill Curation – monitors downstream utility and suppresses skills that cause negative transfer.
Through this workflow–skill–workflow feedback loop, FlowEvo enables agents to accumulate and refine task-solving capability over time without updating model parameters.
Experiments on benchmarks spanning interactive environments (ALFWorld) and code/math generation (HumanEval, GSM8K) show that FlowEvo achieves the best accuracy-cost tradeoff among evaluated baselines under our implementation settings. On ALFWorld, FlowEvo achieves an 82.8% success rate—23.6 percentage points above the strongest baseline—while its average token usage per episode is less than half that of the most efficient baseline. Controlled ablations confirm that each mechanism contributes to the overall result.
Code: https://github.com/DEFENSE-SEU/FlowEvo
Subjects: Artificial Intelligence (cs.AI)
Cite as: arXiv:2607.21596 [cs.AI]
via ArXiv AI