Enhanced ChatGPT Clone: Features Agents, MCP, Skills, DeepSeek, Anthropic, AWS, OpenAI, Responses API, Azure, Groq, o1, GPT-5, Mistral, OpenRouter, Vertex AI, Gemini, Artifacts, AI model switching, message search, Code Interpreter, langchain, DALL-E-3, OpenAPI Actions, Functions, Secure Multi-User Auth, Presets, open-source for self-hosting. Active
TypeScript aianthropicartifactsawsazurechatgptchatgpt-cloneclaudeclonedeepseekgeminigooglegpt-5librechatmcpo1openairesponses-apivisionwebui
const MAX_TOOLS = 256;
const NAME_MAX_LENGTH = 256;
type Call = { name?: string | null };
function parse(input: unknown): unknown {
if (typeof input !== 'string') {
return input;
}
try {
return JSON.parse(input);
} catch {
return undefined;
}
}
/**
* The tools a round called, from the round's own input. The SDK writes that
* input as the list of calls the round ran, but leaves the whole graph state in
* place when it finds none, so only a list whose every entry names a tool is
* read, and only the names leave it: arguments and anything else recorded there
* stay behind the deployment's content setting. The names are exact or absent: a
* client checks them against the chat's own record of the round, which a
* shortened list or a shortened name would fail for a round that matches.
*/
export function toolRoundNames(input: unknown): string[] | undefined {
const calls = parse(input);
if (!Array.isArray(calls) || calls.length === 0 || calls.length > MAX_TOOLS) {
return undefined;
}
const names: string[] = [];
for (const call of calls as Array<Call | null>) {
const name = call != null && typeof call === 'object' ? call.name : undefined;
if (typeof name !== 'string' || name === '' || name.length > NAME_MAX_LENGTH) {
return undefined;
}
names.push(name);
}
return names;
}