@bitkyc08/opencodex
Version:
Universal provider proxy for OpenAI Codex & Claude Code — use any LLM with Codex CLI/App/SDK and Claude Code
31 lines (29 loc) • 1.48 kB
text/typescript
/** Existing model-management commands; use the exact ID from `live`, including native IDs. */
export function modelSelectionNextSteps(provider: string, afterLogin = false) {
const name = provider === "codex" || provider === "chatgpt" ? "openai" : provider;
return {
provider: name,
afterLogin,
requiresRunningProxy: true,
commands: {
list: `ocx models live --provider ${name}`,
enable: 'ocx models enable "<model-id-from-list>"',
disable: 'ocx models disable "<model-id-from-list>"',
enableNative: 'ocx models enable "<model-id-from-list>" --native',
disableNative: 'ocx models disable "<model-id-from-list>" --native',
enableAll: `ocx models provider ${name} on`,
disableAll: `ocx models provider ${name} off`,
},
};
}
export function modelSelectionGuidance(provider: string, afterLogin = false): string[] {
const next = modelSelectionNextSteps(provider, afterLogin);
return [
afterLogin ? "After login completes, manage model switches with:" : "Manage model switches (the provider stays active):",
" Start the proxy first if needed: ocx start",
" Replace <model-id-from-list> with an exact ID printed by the list command.",
" For rows marked native, use the --native variants (including IDs containing /).",
...Object.values(next.commands).map(command => ` ${command}`),
" If initial discovery is still pending, check the provider connection and retry: ocx sync",
];
}