AGENT SKILL

builtin-envs

Choose and troubleshoot Gymnasium built-in environment families, optional extras, registry IDs, action masks, and render modes.

License 许可
MIT
最近更新
2026-09-03
Stars
★ 351
Forks
27

01 前言 Front Matter

name
builtin-envs
license
MIT
metadata
{ "disco-role": "operating" }
description
Choose and troubleshoot Gymnasium built-in environment families, optional extras, registry IDs, action masks, and render modes.
disable-model-invocation
true

02 资源文件 5

03 SKILL.md 正文

Gymnasium Built-in Environments

Use this sub-skill when a task involves selecting or using Gymnasium's built-in environment families, deciding which optional extra is needed, inspecting registered environment IDs, handling Toy Text action masks, or diagnosing backend/render/plugin failures.

Start Here

  • Choose the smallest environment family and optional extra that matches the task; avoid gymnasium[all] unless a disposable environment explicitly needs every backend. See references/optional-dependencies.md.
  • Inspect availability before installing anything: use gym.spec("CartPole-v1"), gym.registry, or gym.pprint_registry(); see references/builtin-environments.md.
  • Use versioned IDs exactly, including namespaces such as phys2d/CartPole-v1 or plugin namespaces such as ALE/...; do not guess unversioned names.
  • For Taxi action masks and render mode choices, use references/action-masking-and-rendering.md and the bundled scripts/action_mask_smoke.py.
  • For missing pygame, Box2D, MuJoCo, ALE/ROMs, shimmy, or media dependencies, use references/troubleshooting.md before broad installs.

Route by Task

Task Read
Pick a built-in family or verify an env ID references/builtin-environments.md
Decide the minimal package extra references/optional-dependencies.md
Use Taxi action masks or render modes references/action-masking-and-rendering.md
Diagnose backend, plugin, ROM, ID, or compatibility errors references/troubleshooting.md
Prove Taxi action-mask support locally scripts/action_mask_smoke.py

Minimum Correct Patterns

import gymnasium as gym

spec = gym.spec("CartPole-v1")
print(spec.id, spec.max_episode_steps, spec.entry_point)

env = gym.make("CartPole-v1")
obs, info = env.reset(seed=123)
obs, reward, terminated, truncated, info = env.step(env.action_space.sample())
env.close()
import gymnasium as gym

env = gym.make("Taxi-v4")
obs, info = env.reset(seed=123)
mask = info.get("action_mask")
action = env.action_space.sample(mask) if mask is not None else env.action_space.sample()
obs, reward, terminated, truncated, info = env.step(action)
env.close()

Scope Boundaries

  • Use ../environment-api/SKILL.md for generic Env loops, custom environment classes, registration, check_env, and old Gym reset/step migration.
  • Use ../vectorization/SKILL.md for gym.make_vec, vectorized built-in environments, batched actions, vector wrappers, and autoreset behavior.
  • Use ../wrappers-recording/SKILL.md for RecordVideo, wrapper chains, recording episode statistics, and media wrapper ordering.
  • Use ../spaces-data/SKILL.md for detailed space constructors, contains, flattening, dtypes, and shape debugging.

Keep runtime guidance self-contained. Do not depend on Gymnasium source, tests, examples, documentation paths, or local checkout paths at runtime.

04 同仓库更多 Skills

VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-triton-opt

Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-solve

END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-skill

PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-pattern-fusion

Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-orchestrate

Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-feedback

Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).

★ 351