AGENT SKILL

capability-discovery

Discover TotalSegmentator tasks, classes, modalities, license flags, and ROI subset names without loading models.

License 许可
Apache-2.0
最近更新
2026-09-03
Stars
★ 351
Forks
27

01 前言 Front Matter

name
capability-discovery
license
Apache 2.0
metadata
{ "disco-role": "operating" }
description
Discover TotalSegmentator tasks, classes, modalities, license flags, and ROI subset names without loading models.
disable-model-invocation
true

02 资源文件 3

03 SKILL.md 正文

Capability Discovery

Use this sub-skill when an agent needs to inspect what TotalSegmentator can segment before planning a run. It owns safe discovery through totalseg_info, totalsegmentator.registry, and the bundled registry helper.

Start Here

  • Use totalseg_info --json for the full machine-readable capability registry: tasks, modality, license flag, and class maps.
  • Use totalseg_info --list-tasks to show every selectable task with modality, license requirement, and class count.
  • Use totalseg_info --classes -ta <task> to list valid class names for one task; these names are the valid --roi_subset values.
  • Use scripts/dump_task_registry.py when a pipeline needs JSON filtering by task, modality, open-license status, or ROI validation.
  • Read references/registry-reference.md for command/API schemas and common planning recipes.
  • Read references/troubleshooting.md when task names, class names, license filters, or discovery commands fail.

Routing Boundaries

Safe Discovery Rules

  • Prefer registry-based discovery over hard-coded task or class names; the registry is the same source used by CLI validation.
  • Treat ROI subset names as exact registry class names for the selected task, not labels from another task and not numeric label ids.
  • Check license_required before selecting a task; this sub-skill can filter licensed tasks, but license acquisition/setup is owned elsewhere.
  • Discovery commands here do not run segmentation, download weights, require a GPU, or load model weights.

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