Feature · Nodes
Turn knowledge into new knowledge.
Nodes is an infinite canvas where memory, files, skills, tools, prompts and any large language model wire together into multi-agent workflows. Drop a model in, connect what it should know, and run.
The short version.
This is where Akhrots stops storing knowledge and starts creating it. Nodes is a visual programming surface with typed sockets, so wires only connect where they make sense. Feed an engine your memory as context, give it a skill and the tools it needs, and chain several models together into a workflow that produces something genuinely new.
What you get.
- Eight node types. Memory, File, Skill, MCP, Prompt, Engine, Output and Note — each with its own colour and role.
- Type-safe wiring. An Engine has four input sockets — Context, Skill, Tools and Prompt — and wires only land where they fit.
- Any model as a node. Drop in a frontier or local model and make it part of the flow.
- Multi-page canvas. Build several workflows side by side as tabbed pages, each remembering its own nodes and view.
- File attachments. Drag PDFs, images, docs and spreadsheets straight onto the canvas as inputs.
- Run the graph. Execute the DAG and watch results appear at the Output nodes.
Under the hood.
13+ models, ready to wire
12 built-in skills
6 tools
From input to insight.
Questions, answered.
Which AI models can I use inside Nodes?
Nodes supports 13+ models — Claude Opus 4.8, Sonnet 4.6, Haiku 4.5 and Fable 5; GPT-5, GPT-5 mini and o4; Gemini 2.5 Pro and Flash; plus Grok 4, Llama 4 Maverick, Mistral Large and DeepSeek V3 — wired with type-safe sockets alongside your skills, tools and memory.
What is an MCP node?
An MCP node connects any Model-Context-Protocol agent or tool into your workflow over an open standard, so Akhrots can use external capabilities the same way it uses its built-in tools.