dat-lequoc/dsh-kiro0

dsh-kiro

Kiro provider for DeepSeek Harness with multi-method login, profile refresh, and live model discovery

AI 分析

核心用途是接入 Kiro 模型服务。适合需要使用 Kiro 平台模型、配置代理及微调推理努力(reasoning effort)的 DSH 用户。

パッケージ
dsh-kiro
バージョン
0.1.3
ライセンス
MIT
最終更新
2026/08/24

インストール

$npx -p @deepseek-ai/dsh dsh plugin --profile web add github:dat-lequoc/dsh-kiro

ドキュメント

README 全文を読む ↗

Configuration

Put machine-level configuration in $DSH_HOME/settings.yaml (normally ~/.dsh/settings.yaml):

llm-kiro:
  proxyUrl: http://proxy.example:1082
  reasoningEffort: medium

Every field is optional:

FieldDefaultMeaning
proxyUrldirectHTTP/HTTPS proxy for Kiro and OIDC requests; credentials in the URL are supported.
regionsigned-in token regionSelects the q..amazonaws.com endpoint.
profileArnaccount defaultCodeWhisperer profile used for requests and model discovery.
thinkingenableddisabled restricts all models to effort off.
reasoningEffortoffDefault: off, low, medium, or high.
defaultContextWindow200000Fallback capacity when discovery reports no exact limit.
modelsbundled fallbackAdvisory fallback catalog; live account discovery normally replaces it.
streamIdleTimeoutMs300000Maximum idle time while a provider read is outstanding.
tokenExpiryBufferMs300000Refresh an access token this long before expiration.
retryPolicybounded normalProvider retry policy executed by dsh-llm-retry.

To pin values to one profile instead, patch the bundle row by id in $DSH_HOME/profiles//cordis.patch.yml:

- id: llm-kiro
  config:
    proxyUrl: http://proxy.example:1082

Do not wrap this override in insert:; the bundle already inserts llm-kiro.