dsh-llm-quota-retry

Keep a DeepSeek Harness agent step alive through an exhausted account quota: after the base provider retry policy gives up on a QUOTA failure, retry the same request once per hour without an attempt limit, and publish the entering, leaving, and current state of that long-term retry to other plugins. 适合需要保证长任务不因临时额度耗尽而中断、需自动重试的无人值守场景。

Package

dsh-llm-quota-retry

Compatibility

Unverified

Version

0.2.0

License

MIT

Last updated

Sep 30, 2026

Configuration

Key	Default	Meaning
`longTermDelayMs`	`3600000`	Interval between long-term attempts, in milliseconds. Must be a positive finite number no greater than `2147483647`.
`defaultEnabled`	`true`	The switch a session starts from before it owns a setting of its own. `false` makes the whole allowance response opt-in.
`pools`	none	Named, ordered route lists a session may be moved between. See below.
`balances`	none	One balance script per provider, reporting what it has left, plus the `apiKeyEnv` credential it runs with. See below.
`failOpen`	`true`	Whether a route nothing reported on counts as usable. `false` makes an unread provider unusable, so only a route a script vouches for is ever chosen.
`exhaustedCooldownMs`	`3600000`	How long a provider stays marked after a request failed with an exhausted allowance. Must be a positive finite number no greater than `2147483647`.

Unknown keys and invalid values throw at activation, so a misconfigured row fails loudly instead of silently retrying on a wrong schedule.

A row that turns the default off makes the allowance response opt-in for the whole deployment; the row below also shortens the interval.

- id: llm-quota-retry
  name: 'dsh-llm-quota-retry'
  config:
    longTermDelayMs: 1800000
    defaultEnabled: false

dsh-llm-quota-retry

Install

README

Configuration