Consider it done. The open-source AI agent that works out of the box · 想到,就能做到。开源、开箱即用的 AI Agent。
TypeScript agentai-agentai-assistantandroidclaude-codecodexelectroniosllmmacosreact-nativetypescriptwindows
import type { UtilityModelTransport } from './utilityModelProfiles';
/** A credential-safe explanation for why one utility-model candidate was not usable. */
export type UtilityTextAttemptReason =
| 'unsupported_transport'
| 'agent_unavailable'
| 'model_unavailable'
| 'not_authenticated'
| 'auth_probe_failed'
| 'api_key_missing'
| 'endpoint_missing'
| 'timeout'
| 'empty_response'
| 'http_error'
| 'request_failed';
/** Fields shared by every candidate considered by the utility-model fallback chain. */
type UtilityTextAttemptBase = {
providerId: string;
model: string;
transport: UtilityModelTransport;
status: 'skipped' | 'failed';
};
/** One candidate diagnostic; HTTP status is valid only for HTTP failures. */
export type UtilityTextAttempt =
| (UtilityTextAttemptBase & {
reason: 'http_error';
/** Safe transport metadata; response bodies are deliberately never exposed. */
httpStatus: number;
})
| (UtilityTextAttemptBase & {
reason: Exclude<UtilityTextAttemptReason, 'http_error'>;
httpStatus?: never;
});
/** Stable high-level failure categories shared by desktop UI and scheduler MCP. */
export type UtilityTextFailureReason =
'no_candidate' | 'all_candidates_failed' | 'empty_response' | 'timeout';
/** Structured diagnostic returned after the configured fallback chain is exhausted. */
export type UtilityTextFailure = {
ok: false;
reason: UtilityTextFailureReason;
attempts: UtilityTextAttempt[];
};
/** Successful utility-model response, including the candidate that produced it. */
export type UtilityTextSuccess = {
ok: true;
text: string;
providerId: string;
model: string;
transport: UtilityModelTransport;
};
/** Result of a utility-model text request. Expected availability failures never collapse to null. */
export type UtilityTextResult = UtilityTextSuccess | UtilityTextFailure;