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
import { InstructionsPromptErrorCode } from 'librechat-data-provider';
import type { AgentInstructionsPrompt, FiltersConfig } from 'librechat-data-provider';
import type {
InstructionsPromptAccess,
InstructionsPromptAccessUser,
InstructionsPromptAccessRequest,
} from './access';
import type { LinkedInstructionsConfig } from './linked';
/** User-safe copy for each stable `instructionsPrompt` write-rejection code. */
const INSTRUCTIONS_PROMPT_ERROR_MESSAGES: Record<InstructionsPromptErrorCode, string> = {
[InstructionsPromptErrorCode.UNAVAILABLE]: 'The selected prompt is not available.',
[InstructionsPromptErrorCode.FORBIDDEN]: 'You do not have access to the selected prompt.',
[InstructionsPromptErrorCode.VALIDATION_FAILED]: 'Unable to validate the linked prompt',
};
export interface InstructionsPromptLinkErrorResponse {
readonly status: 400 | 403 | 500;
readonly body: { readonly error: string; readonly code: InstructionsPromptErrorCode };
}
/**
* Builds the HTTP-shaped `{ status, body }` response for a stable `instructionsPrompt`
* error code, pairing it with its one approved, user-safe message. Shared by every
* caller that maps a stable code to a response, so the same code always carries the
* same copy: the per-operation rejection below, and `checkInstructionsPromptWrite`'s
* own unexpected-failure (VALIDATION_FAILED) case.
*/
export function buildInstructionsPromptError(
status: InstructionsPromptLinkErrorResponse['status'],
code: InstructionsPromptErrorCode,
): InstructionsPromptLinkErrorResponse {
return { status, body: { error: INSTRUCTIONS_PROMPT_ERROR_MESSAGES[code], code } };
}
/**
* Validates a create/update write of `instructionsPrompt` against the agent's stored
* link, before any write lands, and maps a rejection to the HTTP boundary's
* `{ status, body }` shape. Returns `null` when the write may proceed. The caller
* (`/api`) owns only the HTTP call; every ok/forbidden/unavailable decision lives in
* `access.validateLinkWrite`.
*/
export async function getInstructionsPromptLinkError({
access,
user,
previous,
next,
filters,
linkedInstructions,
req,
}: {
access: Pick<InstructionsPromptAccess, 'validateLinkWrite'>;
user: InstructionsPromptAccessUser;
previous: AgentInstructionsPrompt | null | undefined;
next: AgentInstructionsPrompt | null | undefined;
filters?: FiltersConfig;
linkedInstructions?: Pick<LinkedInstructionsConfig, 'langfuse'>;
/** Forwarded unexamined to `validateLinkWrite` so its role lookup can reuse the
* caller's per-request role cache. */
req?: InstructionsPromptAccessRequest;
}): Promise<InstructionsPromptLinkErrorResponse | null> {
const result = await access.validateLinkWrite({
user,
previous,
next,
filters,
linkedInstructions,
req,
});
if (result.ok) {
return null;
}
return buildInstructionsPromptError(result.status, result.code);
}