State is hidden
You know an agent is running. You cannot tell what it is blocked on without opening every tab.
Orchestration for coding agents
Wiregraph turns concurrent coding-agent runs into a live graph you can inspect, steer, fork, and resume—without replacing your terminal or your agents.
For developers running several agent tasks at once.
The problem
You know an agent is running. You cannot tell what it is blocked on without opening every tab.
The reasoning behind a change is stranded in a scrollback buffer nobody will revisit.
Alternative approaches become ad hoc branches with no visible relationship to the original work.
The interaction model
A node is a unit of work with its own context, terminal, changes, and history. Follow the active path, intervene where needed, and preserve the route that produced the result.
Focused capabilities
Every agent task has a visible state, branch, owner, and path through the work.
Inspect the commands, tool calls, output, and changes behind any node without switching context.
Try another approach from a decision point while the original run stays intact and inspectable.
Return to the decisions and outputs that produced a result instead of reconstructing them from chat logs.
How it fits
Wiregraph is the orchestration and observability layer around your existing workflow. It does not replace the coding agent, shell, repository, or review process.
Early access
Wiregraph is early and built independently. The first group is for developers already coordinating multiple coding-agent tasks and willing to give direct feedback.