Checkout rewrite epic — CHK-210
The scope and acceptance criteria.
A single feature spans the spec, the threads that shaped it, the tickets that scoped it, and the PRs that shipped it. Virgo assembles that connected history on demand. You start with the whole picture instead of reconstructing it tool by tool.
One checkout rewrite lives in the spec, the threads that shaped it, the tickets that scoped it, and the PRs that shipped it. To understand it, you open every tool and rebuild the picture by hand.
The scope and acceptance criteria.
Where recalc-at-capture was decided.
Dropped the second tax call. Reverted in canary.
Every piece is here. It is just spread across six tools.
Cart → address → capture, and what each step owns.
Where checkout, tax, and payments meet.
Where the rewrite was scoped.
Virgo connects these sources into one reasoning graph. When you ask about the checkout rewrite, it queries each one, pulls the piece it holds, and returns the connected context. You stop reconciling tabs by hand.
Behind one question, Virgo walks the reasoning graph and returns the connected context behind it, already assembled.
The engineer reads the assembled change context while the agent acts on it.
The spec, the discussion, the tickets, and the code arrive connected. You see how a change came to be and why it works the way it does before you touch it.
An agent handed a snippet writes against the snippet. With the assembled context (the decision, the constraint, the reverted attempt), it reasons about the real change. Virgo retrieves across the graph at query time, over MCP, for any connected agent.
Early evaluations reached 97% context retrieval accuracy, compared with about 68% for baseline RAG or hybrid search.