AGENT SKILL

hitl-apps-and-interaction

Reason about Habitat-HITL realtime apps, viewers, interaction services, networking, policy integration, and graphics troubleshooting.

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

01 前言 Front Matter

name
hitl-apps-and-interaction
license
MIT
metadata
{ "disco-role": "operating" }
description
Reason about Habitat-HITL realtime apps, viewers, interaction services, networking, policy integration, and graphics troubleshooting.
disable-model-invocation
true

02 资源文件 4

03 SKILL.md 正文

Habitat-HITL Apps and Interaction

Use this sub-skill when a task involves Habitat-HITL realtime human interaction applications: desktop viewers, VR or web clients, app/session services, GUI input and drawing, remote-client messages, controller helpers, policy inference inside interactive loops, or graphics/network troubleshooting.

Start Here

  • Read HITL workflows to choose between a safe probe, a headed desktop app, a headless service, a web/VR server, or a custom AppState.
  • Read API reference for hitl_main, AppState, AppService, input/drawing/message helpers, controller helpers, and policy boundaries.
  • Read Troubleshooting before launching full apps on headless, remote, containerized, or graphics-constrained systems.
  • Run hitl_config_probe.py for import/config checks that avoid opening windows, creating Habitat environments, or starting network servers.

Routing Rules

  • Use this sub-skill for habitat_hitl, hitl_main, GUI window/input/drawer/text/UI helpers, remote-client/websocket services, VR interaction, and HITL app state/session structure.
  • Use tasks-datasets-and-envs for core Habitat Env, task, dataset, simulator, and episode semantics outside realtime HITL app orchestration.
  • Use baselines-training-and-evaluation for training, evaluation jobs, checkpoints, PPO/DDPPO configuration, and non-HITL policy workflows.
  • Use setup-and-configuration for installation, package extras, Habitat-Sim build choices, data download basics, and Hydra configuration fundamentals.
  • Use extension-patterns for general extension authoring patterns that are not specifically HITL app interaction.

Safe Operating Posture

  • Prefer the bundled probe script when the user asks whether HITL is installed, whether a module imports, or whether a config looks like a HITL config.
  • Treat full viewers, rearrange apps, VR clients, and websocket servers as side-effectful: they may need a display, OpenGL/EGL support, Habitat-Sim with Bullet, datasets, free ports, and sometimes baseline policy dependencies.
  • Do not call hitl_main as a diagnostic unless the user explicitly wants to launch a HITL runtime and the environment is prepared for graphics or a deliberate headless service run.
  • Keep app-specific launch commands in the user's working project context; this skill provides decision logic and safe probes rather than bundled full viewer or VR launchers.

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