3dgs-engineering-guide
Skill186 starsChanged 57 days agoApache-2.0View on GitHub
Guide for deploying 3DGS from research to production: 10 industry verticals, engineering stack, GIS
In this repository19 other files it gives its agents.
- awesome-gaussian-skillsCLAUDE.md
- awesome-gaussian-skillsSkill
- 3dgs-articulated-reasonerSkill
- 3dgs-code-reviewerSkill
- 3dgs-compression-deploySkill
- 3dgs-experiment-plannerSkill
- 3dgs-mcp-rendererSkill
- 3dgs-method-compareSkill
- 3dgs-paper-readerSkill
- 3dgs-spatial-agentSkill
- 3dgs-training-debuggerSkill
- 3dgs-visualizerSkill
- cad-mesh-3dgsSkill
- cg-paper-writingSkill
- nerf-to-3dgs-migratorSkill
- ontologySkill
- patent-software-ipSkill
- research-expert-teamSkill
- super-agentSkill
The file
We split this skill into the rules it gives an agent
10 rules. About 0 are common, 9 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 63
If the user does not specify, defaults are: domain=all, focus=all.
- Distinctive: line 68
Read these files from static/:
- Very specific: line 72
All domain values load static/industry-landscape.md (the landscape fragment covers all 10 verticals + world models).
- Distinctive: line 92
After loading the required fragments:
- Distinctive: line 114
No invented data: Never fabricate deployment metrics, VRAM numbers, or system requirements not in the loaded reference files.
- Distinctive: line 115
No hallucinated citations: Never invent paper titles, authors, DOIs, arXiv IDs, or venue names.
- Distinctive: line 116
No silent speculation: If you are uncertain about a technical detail, explicitly flag it with "[UNCERTAIN]" rather than presenting it as …
- Distinctive: line 117
No method misattribution: Do not assign features, results, or mechanisms from one method to another.
- Distinctive: line 118
No oversimplified recommendations: Do not reduce complex deployment trade-offs to a single "use this" judgment without considering the …
- Distinctive: line 131
If you cannot find a method, pattern, or data point in the loaded files, say so explicitly.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
3DGS Engineering Guide (Router)
Architecture: Axis-driven static/dynamic router. Do NOT try to apply the engineering guidance from memory or from this router alone. Always load fragments from disk as described below.
Step 1 — Detect Request Axes
Analyze the user's request to determine axis values:
Axis: domain
| User Intent | domain value |
| Autonomous driving simulation, sensor sim | autonomous-driving |
| Digital twin, smart city, GIS | digital-twin |
| Cultural heritage, museum, 3D archiving | culture |
| Film, game production, virtual production | film-games |
| E-commerce, 3D product display, WebAR | ecommerce |
| Industrial inspection, drone survey | industrial |
| AR/VR/MR headsets, spatial computing | ar-vr |
| BIM, architecture, as-built verification | bim |
| Robotics, embodied AI, manipulation | robotics |
| Military simulation, defense | military |
| World model, interactive 3D world | world-model |
| Unspecified or broad engineering question | all |
Axis: focus
| User Intent | focus value |
| Data acquisition, reconstruction, deployment pipeline | pipeline |
| Quality assurance, scalability, cross-platform | best-practices |
| Debugging artifacts, OOM, deployment failures | pitfalls |
| Choosing tools by use case/platform/scale | decision-tree |
| Specific technology stack info (formats, compression, engines) | tech-stack |
| GIS integration, spatial analysis, S3M/Cesium | gis-toolchain |
| Unspecified or comprehensive engineering guidance | all |
If the user does not specify, defaults are: domain=all, focus=all.
Step 2 — Load Required Fragments
Always Load (every invocation)
Read these files from static/:
- static/core-stance.md — Role, workflow, deployment scale reference, terminology, guardrails
On-Demand Load (by detected domain)
All domain values load static/industry-landscape.md (the landscape fragment covers all 10 verticals + world models). The Router will focus attention on the relevant section within the loaded fragment.
On-Demand Load (by detected focus)
| focus | Fragment(s) to Load |
| pipeline | static/tech-stack.md |
| best-practices | static/best-practices.md |
| pitfalls | static/pitfalls.md |
| decision-tree | static/decision-trees.md |
| tech-stack | static/tech-stack.md |
| gis-toolchain | static/gis-toolchain.md |
| all | All focus fragments (tech-stack, best-practices, pitfalls, decision-trees, gis-toolchain) |
Reference Load (when citing specific methods)
- static/reference-papers.md — Method tables organized by domain
Optimization: For a focused question (e.g., "how to deploy 3DGS on Quest 3"), load core-stance + industry-landscape (domain=ar-vr section) + tech-stack. For broad planning (e.g., "start a digital twin project"), load all fragments.
Step 3 — Execute Engineering Guidance
After loading the required fragments:
- Identify use case from domain axis — focus on the relevant industry section
- Recommend pipeline from tech-stack fragment
- Reference specific methods from reference-papers fragment and knowledge base
- Provide concrete next steps — actionable items, not generic advice
- Warn about pitfalls from pitfalls fragment — highlight domain-specific failure modes
Cross-Skill Routing
This skill focuses on production engineering. For related workflows:
- Method comparison → 3dgs-method-compare (for selecting methods for deployment)
- Code review/bug detection → 3dgs-code-reviewer (for detecting deployment-critical bugs)
- MCP rendering → 3dgs-mcp-renderer (for real-time rendering integration)
- Spatial intelligence → 3dgs-spatial-agent (for agent-driven deployment scenarios)
- CAD/Mesh integration → cad-mesh-3dgs (for BIM/CAD workflows)
Red Lines
The following are categorical prohibitions. Violating any of these invalidates the output:
- No invented data: Never fabricate deployment metrics, VRAM numbers, or system requirements not in the loaded reference files. If a value is not found, write "data not available" or "N/A".
- No hallucinated citations: Never invent paper titles, authors, DOIs, arXiv IDs, or venue names. Only reference works explicitly present in the skill's knowledge base or provided by the user.
- No silent speculation: If you are uncertain about a technical detail, explicitly flag it with "[UNCERTAIN]" rather than presenting it as fact.
- No method misattribution: Do not assign features, results, or mechanisms from one method to another. Each method's data is specific to that method.
- No oversimplified recommendations: Do not reduce complex deployment trade-offs to a single "use this" judgment without considering the target environment.
Related Skills
- 3dgs-method-compare — Method comparison (use for selecting methods for deployment)
- 3dgs-code-reviewer — Code review (use for detecting deployment-critical bugs)
- 3dgs-mcp-renderer — MCP rendering protocol (use for real-time rendering integration)
- 3dgs-spatial-agent — Spatial intelligence (use for agent-driven deployment scenarios)
Guardrail: Do Not Apply From Memory
Do NOT try to apply the logic, method data, bug patterns, or technical details described in this skill from memory. Always read the SKILL.md and referenced files from disk before producing any output. The knowledge base is updated frequently; stale memory may produce outdated, inaccurate, or fabricated results.
If you cannot find a method, pattern, or data point in the loaded files, say so explicitly. Never invent metrics, venue acceptances, bug patterns, or technical features not present in the source data.
If you like it, please star this repo https://github.com/jaccen/Awesome-Gaussian-Skills
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

