neva: Skill for Codex

Neva to Mermaid

Use this skill to convert any Neva source code into a Mermaid flowchart.

Instructions

1. Read the Neva code to be converted.

2. Generate a Mermaid flowchart TB diagram using the following rules:

   • Layout: Always include this header at the top:

     ---
     config:
       layout: elk
     ---
     flowchart TB
     

   • Ports:

     • Use stadium shapes for external ports :in / :out:
       • ([":data"]), ([":res"]), ([":err"]), etc.
     • In edges, label with the port name (data, res, err, sig, case[0], etc).

   • Components:

     • Use unique Mermaid node IDs.
     • Render component nodes as regular boxes using only their declared instance name: id["name"].
     • Do not annotate component nodes with type names, generic parameters, or usage information.

   • Literals:

     • Render numeric and string literals as rectangles using Mermaid shape syntax:
       • zero@{ label: "0", shape: rect }
       • dot@{ label: "'.'", shape: rect }

   • Connections:

     • Use -- <port> --> labels for the wire’s port name.
     • For a:out -> b:in, label the edge with the output port name (out), unless Neva explicitly uses the input port name in the wiring (e.g., b:sig).
     • Only label an edge with a port name when the Neva source explicitly names a port on that connection.
       • Examples that MUST be labeled:
       • x:res -> y → label edge as res
       • x -> y:sig → label edge as sig
       • x:case[0] -> y → label edge as case[0]

   • If a port name is omitted in Neva, do not infer or guess it. Render an unlabeled edge:

     • x -> y → x --> y

   • Fan-out / Fan-in (IMPORTANT):

     • Mermaid edges do not create real “ports as objects”, so list-wiring must be represented explicitly with virtual junction nodes.
     • A junction node is a small circle: jX(( )) where jX is a unique ID.

     Fan-out: src -> [a, b, :out]

     • Insert a junction node jX(( )).
     • Wire: src -- <port> --> jX
     • Then wire: jX --> a, jX --> b, jX --> out_port