[Feature] 希望能有一个无限重试的选项

[sheaxiang]

Problem / 问题

现在很多中转站都不稳定,有些过几个小时才能恢复,希望增加一个无限重试功能,这样中转站恢复的时候能继续执行,不用人工再去点继续

Proposed change / 期望改动

现在很多中转站都不稳定,有些过几个小时才能恢复,希望增加一个无限重试功能,这样中转站恢复的时候能继续执行,不用人工再去点继续

Alternatives / 其他方案

No response

Additional context / 补充信息

No response

[ZxlDragonDoctor]

I'd like to take this. Proposed scope:

• Add an opt-in "retry until success" mode for transient provider failures (unstable relays that recover after hours).
• Default stays the current bounded budget (ADR 0206: ten retries).
• Infinite mode draws from the same transient / rate-limit classification and backoff schedule; only the stop condition changes.
• Surface retry count and last error in the existing progress status so a long wait is visible.
• Unit tests for: bounded default unchanged; infinite mode keeps retrying after budget exhaustion; cancel/stop still works.

Non-goals: changing the bounded default, nested pi-ai retries, or new UI beyond the existing retry status row.

[vastsa]

已合入：设置 → AI → 默认项 →「无尽重试」。默认关闭；打开后网络/临时故障会一直重试到成功，随时还能点停止。

#759