Where Autonomy Belongs: A Separation of Powers for Agentic Program Evolution
Organizations: Huawei Taylor Lab · Shanghai University of Finance and Economics
Abstract
LLM-driven program evolution rests on three decisions: govern directs the search from accumulated evidence, propose constructs the next candidate, and judge settles validity and fitness. All three appear in every system, and existing work splits into two lines over who makes each. Controller-centric systems strengthen govern and neglect the autonomy of propose, which caps what evolution can reach. Agent-centric systems let a coding agent take over both govern and propose, at a cognitive and computational cost. Neither line protects the independent authority of judge, so benchmark reliability suffers. These failures share one root: autonomy is misallocated across the three decisions. We address this problem with EvoTrias, a separation-of-powers architecture in which each decision has one owner: Structured Governance for what spans evolution rounds, Open Proposal for construction inside a task, and Protected Judgment behind a private evaluator. These owners exchange typed contracts only, and runtime hooks reinforce the boundaries. This separation also lets EvoTrias run as a plugin that turns an existing coding agent into an evolution system. EvoTrias sets a new state of the art on mathematical, systems, and algorithm-design benchmarks. Three further results show what each owner buys: separating Governance from Proposal cuts token cost by 64%, Open Proposal extends evolution to a 680K-line industrial repository, and Protected Judgment keeps benchmark scores faithful to transfer.
Figures & tables
| GLM-5.2 | Opus-5 | |||||||||||
| Task | OpenE. | EvoX | AdaE. | CORAL-1 | CORAL-4 | EvoTrias | OpenE. | EvoX | AdaE. | CORAL-1 | CORAL-4 | EvoTrias |
| Math | ||||||||||||
| Circle-Pack. (Sq.) | 2.54317 | 2.60956 | 2.63152 | 2.62337 | 2.63418 | 2.63484 | 2.63598 | 2.63598 | 2.63598 | 2.63598 | 2.63598 | 2.63599 |
| Circle-Pack. (Rect.) | 2.35930 | 2.34876 | 2.36487 | 2.32097 | 2.36459 | 2.36493 | 2.36583 | 2.36583 | 2.36584 | 2.36583 | 2.36583 | 2.36584 |
| Heilbronn-Tri. | 0.0253 | 0.0302 | 0.0318 | 0.0320 | 0.0332 | 0.0335 | 0.0357 | 0.0361 | 0.0364 | 0.0364 | 0.0365 | 0.0365 |
| Heilbronn-Conv. | 0.0250 | 0.0269 | 0.0275 | 0.0284 | 0.0283 | 0.0302 | 0.0290 | 0.0307 | 0.0304 | 0.0309 | 0.0309 | 0.0309 |
| Configuration | Circle Packing | Signal Processing |
| Full EvoTrias | 2.6348 0.0019 | 0.7329 0.0058 |
| Level 1: Governance and initialization | ||
| w/o Notes | 2.6122 0.0264 | 0.7193 0.0281 |
| w/o Meta-guidance | 2.6087 0.0102 | 0.7041 0.0094 |
| w/o adaptation ( ) | 2.6267 0.0079 | 0.6926 0.0132 |
| Initial directions | 2.5969 0.0448 | 0.6763 0.0554 |
Appendix figures & tables18 assets
Supplementary material from the paper’s appendix.
Appendix
| State | Entered when | Event source |
|---|---|---|
| created | Governance mints identifiers at a convergence check | control-plane call |
| pending | the hook allocates one to an observed executor spawn | before the tool call |
| queued | the spawn returns an agent handle | after the tool call |
| running | the subagent starts | subagent lifecycle |
| completed | the subagent stops with a successful attempt on record | subagent lifecycle |
| failed | the subagent stops with no successful attempt on record | subagent lifecycle |
| Role | Claude Code | OpenCode |
|---|---|---|
| spawn observed | PreToolUse on the agent tool | tool.execute.before on the task tool |
| handle returned | PostToolUse | tool.execute.after |
| subagent starts | SubagentStart | session.created carrying a parent |
| subagent ends | SubagentStop | session.idle |
| Task | Reported quantity | Internal fitness |
|---|---|---|
| Math | ||
| Circle Packing (square) | sum of 26 radii | |
| Circle Packing (rectangle) | sum of 21 radii | |
| Heilbronn Triangles | normalized minimum triangle area | |
| Heilbronn Convex | normalized minimum triangle area | |
| MinMax Distance | squared min-to-max distance ratio | |
| Suite | OpenEvolve | EvoX | AdaEvolve | EvoTrias |
|---|---|---|---|---|
| Math ( runs) | 55.47 | 57.81 | 58.76 | 55.81 |
| ADRS ( runs) | 57.60 | 57.40 | 54.53 | 52.87 |
| Frontier-CS ( runs) | 61.13 | 56.37 | 55.13 | 56.80 |
| Problem | GLM-5.2 (h) | Opus-5 (h) |
|---|---|---|
| Math | ||
| Circle Packing Square | 1.87 | 2.67 |
| Circle Packing Rectangle | 1.84 | 2.11 |
| Heilbronn Triangles | 1.81 | 1.90 |
| Heilbronn Convex | 2.54 | 3.16 |
| MinMax Distance | 3.04 | 3.09 |
| Baseline | Rel. gain | 95% CI | W/T/L | Exact | Holm |
|---|---|---|---|---|---|
| OpenEvolve | +6.403% | 12/0/0 | 0.000488 | 0.00244 | |
| EvoX | +3.010% | 12/0/0 | 0.000488 | 0.00244 | |
| AdaEvolve | +2.047% | 12/0/0 | 0.000488 | 0.00244 | |
| CORAL-1 | +1.704% | 12/0/0 | 0.000488 | 0.00244 | |
| CORAL-4 | +0.640% | 9/0/3 | 0.1646 | 0.1646 |
| Baseline | Mean | 95% CI | W/T/L | Paired | Holm |
|---|---|---|---|---|---|
| OpenEvolve | +9.44 | 14/10/6 | 0.00348 | 0.01045 | |
| EvoX | +8.76 | 11/11/8 | 0.03377 | 0.06755 | |
| AdaEvolve | +10.13 | 13/12/5 | 0.000866 | 0.00433 | |
| CORAL-1 | +8.50 | 15/13/2 | 0.00247 | 0.00989 | |
| CORAL-4 | +3.49 | 10/12/8 | 0.1310 | 0.1310 |
| Method | Circle Packing (Square) | Signal Processing | Heilbronn Triangles |
|---|---|---|---|
| EvoTrias (reference) | 2.6348 0.0019 | 0.7329 0.0058 | 0.0335 0.0022 |
| OpenEvolve | 2.5432 0.0030 | 0.6373 0.0131 | 0.0253 0.0012 |
| agentic proposer | 2.6178 0.0267 | 0.6804 0.0140 | 0.0304 0.0019 |
| vs. single-call | 2.93% | 6.76% | 20.12% |
| vs. EvoTrias | 0.65% | 7.17% | 9.06% |
| AdaEvolve | 2.6315 0.0024 | 0.6961 0.0226 | 0.0318 0.0013 |
| Directory | Lines | Contents |
|---|---|---|
| sat/ | 234K | CP-SAT solver |
| math_opt/ | 130K | unified modeling layer across solvers |
| constraint_solver/ | 108K | legacy constraint programming and routing |
| graph/ | 37K | shortest path, flow, and assignment |
| linear_solver/ | 31K | linear solver wrappers |
| glop/ | 20K | simplex LP solver |
| Dimension | Original evaluator | IID held-out | OOD held-out |
|---|---|---|---|
| Cases | 5 | 25 | 20 |
| Signal families | Five fixed families, one case each | Same five families, five cases each | Five widened families plus five new families, two cases each |
| Length | 500, 600, 700, 800, 900 | 529–997 | 457–1217 |
| Family–length relation | One-to-one; length identifies the case | Multiple lengths per family | Length is decoupled from family |
| Window size | 20 | 20 | 9–67 |
| Time span | 0–10 | 8–14 | 6.5–17 |