Ask an AI assistant about TrussLab

What this page is for#

TrussLab is intentionally different from ordinary CAD. The hard part for a new user is often not finding a button; it is understanding the model: logical nodes and links first, stock members for length, printed attachments and joints for local complexity.

This page gives general-purpose AI assistants a compact operating manual. Copy the context, paste it into ChatGPT, Claude, Codex, or another assistant, then ask practical questions about what you want to model.

The same context is also available as plain Markdown at agent-context.md, and agents that look for site indexes can start from llms.txt.

Copy the context#

Use the copy button near the top of this page when you want to bring an assistant into a TrussLab session. The copied text tells the assistant what TrussLab is, what it can and cannot do, how presets, autoplay, transaction history, and reconciliation work, and how to answer modeling questions responsibly.

After pasting it into your assistant, ask for a concrete workflow rather than a generic explanation. Good questions mention the object, scale, stock material, printed material, expected loads, mounting constraints, and whether you want to start from a preset.

Good questions to ask#

• How would I model a small wall-mounted camera arm from stock tube and printed joints?
• Which preset is closest to a lightweight drone frame, and what should I change first?
• I want to make a rectangular fixture frame with diagonal bracing. What TrussLab concepts do I need?
• When should I use a physical link attachment instead of a virtual link attachment?
• How should I think about reconciliation before exporting parts for printing?
• How can I use the transaction log to try full-frame reconciliation safely?
• What should I verify before treating this exported model as build-ready?

How assistants should answer#

An assistant should explain TrussLab in terms of the editor's real concepts: nodes, logical links, physical links, attachments, attachment connections, attachment offsets and end pinning, reconciliation, transaction log, presets, autoplay, properties, and export.

For modeling questions, the assistant should first classify the task. Is the user trying to design the overall frame, a printable attachment, a connected joint, or a finished physical assembly? Then it should propose a short sequence of editor actions and name the parts of the model that should stay parametric.

When a preset is close, the assistant should recommend starting there, watching or scrubbing the narrated autoplay, taking over at the relevant step, and changing properties before moving individual geometry.

History-aware guidance#

Assistants should treat the transaction log as a normal part of the TrussLab workflow. Users can try complicated modeling actions, inspect the result, and use Undo/Redo or return to an earlier history step if the change is not better.

This is especially useful for risky-looking operations such as full-frame reconciliation, changing a joint strategy, moving attachments, or sweeping a parameter through several values. The right advice is often: try the operation, inspect the generated links, joints, and clearances, then use Undo/Redo or continue from an earlier history step.

Limits and verification#

AI assistance is useful for orientation and workflow planning, but it is not a structural certification tool. TrussLab does not currently replace engineering judgment, physical testing, load calculations, material choice, printer process validation, or inspection of exported geometry.

Before building anything that carries meaningful load, verify the design yourself. Check member lengths, clearances, printed part strength, fasteners, mounting surfaces, print orientation, material behavior, and the consequences of failure.