Pythonic config generation for Hydra (hydra-zen). Eliminates hand-written YAML by generating structured dataclass configs directly from Python objects and functions. Provides make_config, builds, zen, store, and launch utilities for configurable, reproducible, and scalable workflows.
SkillScore100Maintained
Automatic
◆AI Engineer pack
Last verified 8/21/2026
What it changes
Passed 22 of 22 test cases with the skill, and 15 of 22 without it.
Without skill68%
With skill100%
Checks it passed that the model alone missed
Calls `make_config`
Passes `hydra_convert="all"`
Calls `builds` with the `Trainer` class
and 7 more in the full report
7 fixed15 passed either way0 still failing0 broke
Pass-rate lift
+32 pp
Tokens
+59%
Agent turns
0%
Response time
-50%
Model gemini-3.6-flashMeasured 8/21/2026v122 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 7 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 7
Prompt
I need an ad hoc typed config object with fields for `learning_rate` and `batch_size`, but I don't want to define a full dataclass manually. Using the Python-first Hydra tool, what function do I call to generate this?
What was checked, without and with the skill
Checked
Without skill
With skill
Calls `make_config`
Missed
Passed
case-07 · gemini-3.6-flash · 8/21/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md206 linessingle file
hydra-zen
Overview
hydra-zen is a Python-first layer on top of Hydra that removes most Hydra-specific boilerplate and eliminates hand-written YAML configs. It dynamically generates structured dataclass configs from functions, classes, and call signatures, then integrates them into Hydra workflows. Use this skill when you want typed, composable, reproducible experiment configuration without maintaining large YAML trees.
When to Use This Skill
This skill should be used when:
You want Hydra benefits without hand-writing YAML configs
You need type-safe, dataclass-driven experiment configuration
You want to generate configs directly from Python call signatures
You want to reduce Hydra boilerplate in ML/research codebases
You need Pythonic launch/store APIs for configurable workflows
You are building reproducible training scripts with complex parameterization
Core Capabilities
1. Installation
pip install hydra-zen
2. builds() — Generate Configs from Callables
from hydra_zen import builds
from torch.optim import AdamW
AdamWConf = builds(AdamW, lr=1e-3, weight_decay=1e-2)
# AdamWConf is a dataclass config that Hydra can instantiate