Frequently Asked Questions

Everything you need to know about Kiln. Can't find your answer? Open an issue on GitHub.

General

What is Kiln?
Kiln is open-source infrastructure that lets AI agents design new objects, discover existing models, and autonomously slice, print, and monitor 3D printers. It exposes 456+ tools through the Model Context Protocol (MCP), so any compatible AI agent — like Claude, GPT, or a custom agent — can run full print workflows with safety guardrails built in. Describe what you want or upload a sketch, and Kiln handles everything from generation to the finished print. Think of it as the intelligence layer between an idea and a physical object.
How is Kiln different from just using OctoPrint or Moonraker?
OctoPrint and Moonraker are great printer management interfaces built for humans. Kiln sits on top of them (and Bambu Lab, Elegoo, and Prusa Link) to provide a unified, AI-friendly interface. It adds safety guardrails, a job queue, fleet management, slicing integration, and a standardized protocol — things you need when an autonomous agent is in the loop, not a person clicking buttons.
How is Kiln different from decentralized manufacturing networks/providers?
Kiln is orchestration infrastructure for agents — it does not operate its own marketplace or manufacturing network. Instead, Kiln searches third-party model marketplaces (Thingiverse, MyMiniFactory, Cults3D) and routes fulfillment orders to third-party providers (like Craftcloud). The provider remains merchant of record for any outsourced orders. See the product boundary page for a direct comparison.
Is Kiln free?
Local printing is free forever — unlimited prints on up to 2 printers, no strings attached. Paid tiers add fleet management, larger queues, priority support, and access to outsourced manufacturing through fulfillment providers. See the pricing page for details.
Is Kiln open source?
Yes. Kiln is fully open source under the MIT license. You can self-host, modify, and contribute. The source code is on GitHub.

Getting Started

What printers does Kiln support?
Kiln supports printers running OctoPrint, Moonraker (Klipper), Bambu Lab (X1C, P1S, P1P, A1, A1 Mini), Elegoo (Centauri Carbon, Saturn, Mars via SDCP; Neptune 4 and OrangeStorm Giga via Moonraker), and Prusa Link (MK4, MK3.9, Mini, XL). If your printer runs one of these, Kiln can control it. Safety profiles cover 30+ printer models with per-model temperature and speed limits.
How do I install Kiln?
Install via pip: pip install kiln3d, then run kiln to start the MCP server. Point your AI agent at the server and you're ready to go. Full instructions are on the install page.
What AI agents work with Kiln?
Any AI agent that supports the Model Context Protocol (MCP) can use Kiln — including Claude (via Claude Desktop or Claude Code), custom agents built with the OpenAI API, and any MCP-compatible client. Kiln also provides a REST API and CLI for non-MCP integrations.
What is MCP? Do I need to understand it?
MCP (Model Context Protocol) is an open standard for connecting AI agents to external tools and data sources. You don't need to understand MCP internals to use Kiln — just configure your AI agent to connect to the Kiln MCP server. If you're building custom agents, the whitepaper covers the protocol details.
Can I use Kiln without AI, just as a CLI or API?
Yes. Kiln includes a full command-line interface with 136 commands and a REST API. You can manage printers, queue jobs, slice files, and run print operations entirely from the terminal or your own scripts — no AI agent required.

Safety & Privacy

Can the AI damage my printer?
Kiln is built around the principle that AI should never have unchecked access to hardware. Every print operation passes through a safety layer that enforces per-printer temperature limits, validates G-code commands, runs pre-flight checks before prints, and blocks dangerous operations. Safety profiles cover 30+ printer models with manufacturer-specified limits.
What safety guardrails are in place?
Kiln enforces multiple layers: per-printer temperature and speed limits from a validated safety profile database, G-code command validation that blocks unsafe operations (raw heater commands, dangerous moves), mandatory pre-flight checks before every print, and structured error handling that never silently fails. The whitepaper covers the full safety architecture.
Does Kiln send my data to the cloud?
No. Kiln runs entirely on your local machine. Your printer data, G-code files, and print history stay on your network. The only external connections are ones you explicitly configure — like third-party fulfillment providers or third-party marketplace search.
Can I restrict which tools the AI agent can access?
Yes. Kiln supports tool tiers (essential, standard, full) that control which MCP tools are exposed to the agent. You can restrict agents to read-only monitoring, allow standard print operations, or grant full access including fleet management and raw G-code — depending on your trust level.

Features & Capabilities

Can I run multiple printers?
Yes. Kiln includes a fleet registry and job queue that support multiple printers. The free tier supports up to 2 printers; paid tiers support unlimited printers with priority scheduling across your fleet.
Does Kiln handle slicing?
Yes. Kiln integrates with PrusaSlicer and OrcaSlicer for local slicing, with bundled slicer profiles optimized per printer model. You can also use pre-sliced G-code files. The AI agent can trigger slicing, adjust settings, and go from STL to print in a single conversation.
Can I order prints from external services through Kiln?
Yes. Kiln routes orders to third-party fulfillment providers — currently Craftcloud, which works out of the box with no API key required. Your AI agent can compare quotes from 150+ third-party print services across FDM, SLA, SLS, MJF, and metal, select materials, and place orders through the same interface used for local printing. Kiln does not operate its own manufacturing network — it orchestrates the workflow while the third-party provider remains merchant of record.
Does Kiln support model marketplaces?
Yes. Kiln searches third-party model marketplaces — Thingiverse, MyMiniFactory, and Cults3D — so your AI agent can discover existing models, download files, and send them to your printer in one workflow. Kiln does not operate its own marketplace; it provides search and download adapters for these external platforms.
Can I automate recurring prints or batch jobs?
Yes. Kiln includes a priority job queue with scheduling, so you can queue up multiple prints with different priorities. The scheduler dispatches jobs automatically as printers become available, and you can build fully automated pipelines using the pre-validated print pipeline system (quick_print, calibrate, benchmark).
Can Kiln generate 3D models from text or sketches?
Yes. Kiln supports text-to-3D generation via Gemini Deep Think, Meshy, Tripo3D, and Stability AI, plus parametric code-as-CAD generation via OpenSCAD with a 14-component catalog (threads, gears, boxes, brackets, etc). You can also upload a napkin sketch for image-to-3D conversion. Providers are auto-discovered from environment variables — just set your API key and Kiln finds it.
What is Design Intelligence?
Design Intelligence is Kiln's engineering knowledge base — 52 materials with mechanical/thermal/chemical properties, 20 proven design patterns (snap-fits, living hinges, threads, press-fits), 70 parametric templates, and structural load estimation. When you describe what you want to print, Kiln uses this knowledge to select the right material, apply proven patterns, and validate that the design will actually work before generating geometry.
What does the Printability Engine do?
Before any model reaches your slicer, Kiln analyzes it across 7 dimensions: overhang detection, thin wall analysis, bridging assessment, bed adhesion estimation, support volume calculation, warping risk, and thermal stress. It scores models 0-100, recommends optimal print orientation, and suggests adhesion settings (brim/raft) based on the geometry, material, and your specific printer.
Can Kiln design multi-part assemblies?
Yes. The assembly module validates clearances between parts, detects joint types (snap-fit, press-fit, threaded, adhesive), checks tolerance stacking, and plans splits for models that exceed your build plate. Free tier supports up to 10 parts per assembly.
How does cost estimation work?
Kiln calculates the cost of any print before it starts — material cost (by weight and filament type), electricity, and time. It also provides smart recommendations: cheaper filament alternatives that maintain strength, speed profiles that save time, and quality tradeoffs. Use it to budget projects or compare local vs outsourced manufacturing costs.

Technical

What's the difference between the litepaper and whitepaper?
The litepaper is a non-technical overview — it explains what Kiln does, why it exists, and who it's for, without diving into implementation details. The whitepaper is the full technical document covering protocol design, architecture, adapter interfaces, safety models, and system internals.
What language is Kiln written in?
Python. Kiln is distributed as two pip-installable packages: kiln3d (the MCP server) and octoprint-cli (standalone CLI tool). Both use modern Python with full type hints, dataclasses, and async support.
Can I build my own printer adapter?
Yes. Kiln uses an abstract adapter interface (PrinterAdapter) that any new printer backend can implement. If your printer has a REST, MQTT, WebSocket, or serial API, you can write an adapter that normalizes it to Kiln's standard interface. See the existing adapters (OctoPrint, Moonraker, Bambu, Elegoo, Prusa Link) as reference implementations.
How do I contribute?
Fork the repo, create a branch, and submit a pull request. Kiln uses pytest for testing (5,000+ tests), Ruff for linting, and follows strict patterns documented in the contributing guide. Issues and feature requests are welcome on GitHub.