luma-material

Original：🇺🇸 English
Translated

Use when a Luma / 拾光 / 拾光智能体 / 拾光工具 agent needs to inspect local material libraries, describe material groups, upload or understand materials, search candidates, or prepare PIP matching inputs.

55installs
Sourcezl007700/luma-cli
Added on

NPX Install

npx skill4agent add zl007700/luma-cli luma-material

Tags

Translated version includes tags in frontmatter
luma-agentmaterial-librarypip-matchingcloud-material-understandinglocal-material-management

SKILL.md Content

View Translation Comparison →

Luma Material

Use this skill for Luma / 拾光 local material libraries, cloud material understanding, material search, and PIP matching preparation.
Read 
../luma-shared/SKILL.md
first for common project and output rules.

When To Use

  • The user provides a local material group, for example 
    data/material_library/groups/vlm_ai
    .
  • Reusable groups should be imported into the default Luma material library at 
    ~/.luma/material-library
    .
  • A workflow needs 
    step4_materials_enriched.json
    or 
    step4_material_matches.json
    .
  • The agent needs to inspect which materials are available before PIP rendering.
  • The user wants to upload or understand a material through the backend.

Local Material Group Flow

List available groups:
bash
luma-cli material library path
luma-cli material group list --output material_groups.json
Import a reusable material group into the default library when needed:
bash
luma-cli material library import ./material_library/groups/vlm_ai --replace
Describe one group into a standard materials file:
bash
luma-cli material group describe vlm_ai --output step4_materials_enriched.json
Search candidate materials before planning:
bash
luma-cli material search --materials step4_materials_enriched.json --query "<script_or_scene_text>" --limit 8 --output step4_material_matches.json

Cloud Understanding

If local metadata is missing or weak, understand useful materials one by one:
bash
luma-cli material understand ./materials/a.jpg --output a.meta.json --descriptor-output a.material.json
luma-cli material merge --materials step4_materials.json --meta ./materials_meta --output step4_materials_enriched.json

PIP Matching

After subtitles are segmented and scene units are ready:
bash
luma-cli pip scene --segments step4_segments.json --output step4_scene_units.json
luma-cli pip match --scenes step4_scene_units.json --materials step4_materials_enriched.json --mode auto --output step4_material_matches.json
Use 
--mode cloud
when semantic matching must be backend-only. Use 
--mode local
only as a deterministic fallback.

Agent Rules

  • Prefer the default material library and 
    material group describe <group_name>
    for local material libraries.
  • Do not upload an entire local library unless the user explicitly asks; upload only materials that need cloud understanding.
  • If matching returns zero inserts, report that no suitable materials were found and continue without PIP.
  • Keep 
    materials.json
    , scene units, matches, and PIP plan as separate artifacts.
  • Do not expose backend object storage implementation details to the user unless a command explicitly requires them.