// agentic ai · live tree visualization
Think in trees. Not in lines.
An agentic chat that shows you every step the AI takes — reasoning nodes, tool calls, skill invocations — as a live expanding tree.
Launch AppBring your own API key — nothing stored on the server.
// agent tree
Every response is a live graph of think nodes, tool calls, collapses, and skill invocations — not a black box.
// multi-provider
Gemini 3.5, GPT-5.5, o3, local Ollama — switch providers in one click. No server-side keys stored.
// streaming
Watch the agent reason token-by-token as the tree builds live. Full depth control and budget limits.
// skills
Sub-agents with their own dedicated tool access and system prompts.
Executes Python, JavaScript, or TypeScript in a secure cloud sandbox and returns output, errors, and generated files.
skill_code_runnerFetches and summarises content from URLs — preserves numbers, names, and dates exactly.
skill_web_researcherCreates formatted Word documents (.docx) — headings, tables, lists, and downloads.
skill_docx_skill// tools
15 built-in tools available to the agent on every request.
// web
Real-time web search across the open internet.
Read any web page or HTTP resource as text.
Send a POST request with a JSON body to any endpoint.
// code
Execute Python 3 code in a secure cloud sandbox.
Execute JavaScript (Node.js) in a sandbox.
Execute TypeScript in a sandbox.
Run any shell command inside the sandbox.
Get the public URL of a running sandbox server.
// files
Read a file from the sandbox filesystem.
Write or create a file in the sandbox.
List files and directories in the sandbox.
Delete a file from the sandbox.
Create a directory in the sandbox.
Download a file from a URL into the sandbox.
// utility
Get the current date and time in any timezone.