hillclimb 0.7.0

Hillclimbing on verifiable rewards.
Use Claude Code and Codex to autonomously search for python programs that optimize a score you define.

Website · Docs · Quickstart · CLI reference · Blog · Discord

Give hillclimb a problem — a folder whose verifier scores a solution.py — and a budget. It runs Claude Code, Codex or pi as operators that draft, debug and improve solutions, scores every candidate through your verifier, and keeps the best.

The harness is fixed. The climber — what to try next and how each attempt is prompted — is a block you can swap, edit and share, so two methods can be compared on the same problem under the same budget.

Quickstart

pip install hillclimb
hillclimb connect claude               # or codex, pi; add --agent dummy to any run for no LLM at all

hillclimb init                         # hillclimb.yaml, problems/, runs/ in this folder
hillclimb problem get heilbronn-11     # the verifier is the problem; description.md is the brief
hillclimb verify heilbronn-11          # score the floor solution

hillclimb run heilbronn-11 --budget 10m
hillclimb watch                        # live coding agents and scores (also: chart, tree)
hillclimb stop --all

The best solution lands in runs/<run-id>/searches/heilbronn-11/best/. The walkthrough goes through each step.

Climbers

A climber is one block of config, in a run spec or as the folder's default in hillclimb.yaml:

climber:
  selector_policy: map-elites     # which candidate to build on next
  operator_policy: greedy         # which operator to use on it
  operators: [draft, debug, improve, crossover.py:Crossover]
  tuner: optuna
  params: {num_drafts: 5}

Every slot takes a prebuilt module, a .py file of your own, or package.module:Class. Presets: greedy, openevolve, gepa. Or compose in Python:

from hillclimb import Budget, Climber, Problem
from hillclimb.policies import Greedy
from hillclimb.selectors import MapElites

problem = Problem("heilbronn-11")
budget = Budget(wall_clock="10m", evaluations=40)
climber = Climber(selector_policy=MapElites(num_drafts=5), operator_policy=Greedy())

climber.search(problem, budget=budget)
climber.best, climber.history, climber.to_frame()

climber.start(...) opens the same search to drive by hand, one climber.step() at a time, and Problem(name, score=my_function, ...) defines a problem from a scoring function. Runnable scripts are in examples/.

Compare two head to head:

hillclimb run heilbronn-11 --climber greedy --climber openevolve --parallel-searches 2
hillclimb experiment report <run-id>

Example problems

Exact, noise-free construction problems, each with its best known value as the target line. hillclimb problem list shows the full catalog.

Define your own: hillclimb problem new my-problem writes a small problem that already runs, for you to edit into yours. Kaggle (MLE-bench) and Einstein Arena come in as benchmark problems.

Learn more

Questions, results, ideas: join the Discord.

License

MIT