Skip to content
Get packs

PufferLib RL Guide

Guides an agent through PufferLib reinforcement-learning envs, vectorization, training and checkpoint review

Source Download skill
Repository
K-Dense-AI/scientific-agent-skills
Stars
55,780
Licence
MIT
Files
15 files

What it does

Once installed, the agent can answer questions about PufferLib, a reinforcement-learning library, and pick the correct version (native 5.0, PyPI 3.0.0, or pinned 4.0) before writing code. It explains the Gymnasium-style environment contract, how to wrap environments, how to run training safely, and how to review checkpoints. It also ships small dependency-free Python checks that run synthetic environments locally to validate a contract or benchmark vectorization, without installing PufferLib.

Great for

  • Developers building or porting reinforcement-learning environments who want a checklist for the step/reset contract
  • Researchers comparing PufferLib versions or reproducing a pinned setup
  • Teams reviewing whether an external environment or checkpoint is safe to load

Skip it if

  • Users who need a turnkey training run: the bundled plan tool does not launch training and the native/PufferLib training examples were not executed in review
  • Anyone wanting guaranteed results from untrusted checkpoints, pickles, ROMs or environment packages, which should be sandboxed and hash-verified first (risk of arbitrary code execution)

Try asking

  • “Validate my Gymnasium environment's reset and step contract with 64 steps and 8 episodes”
  • “Which PufferLib version should I use for a C/CUDA environment on a CPU-only machine?”
  • “Wrap my PettingZoo Parallel environment with PufferLib 3.0.0 emulation”
  • “Benchmark vectorization on synthetic environments with the serial backend”
Step-by-step guide: how to use PufferLib RL GuideInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 15 original files, the MIT licence and an attribution note.

    pufferlib.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip pufferlib.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip pufferlib.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Start with the bundled synthetic CLIs; they are CPU-only, bounded, and need no network.
  • Pin the exact PyPI sdist hash or git commit rather than branch names before installing.
  • Build and first-run any native environment in a sandbox without credentials or sensitive mounts.

What's inside

  • references/environments.md
  • references/integration.md
  • references/native-5.md
  • references/policies.md
  • references/training.md
  • references/vectorization.md
  • scripts/__init__.py
  • scripts/_common.py
  • scripts/benchmark_vectorization.py
  • scripts/env_contract_validator.py
  • scripts/env_template.py
  • scripts/inspect_checkpoint.py
  • scripts/repro_plan.py
  • scripts/train_template.py
  • SKILL.md

Redistributed unmodified from K-Dense-AI/scientific-agent-skills under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.