HUSTforever/dsh-thinking-summary ↗★ 0
dsh-thinking-summary
Codex-style per-step reasoning summaries inside DeepSeek Harness's native thinking disclosure. 适合希望直观、快速地阅读和理解AI冗长思考过程的用户。
Other repositories with this package name
Install
$
npx -p @deepseek-ai/dsh dsh plugin --profile web add github:HUSTforever/dsh-thinking-summaryREADME
Read the full README ↗Configuration
| Key | Default | Meaning |
|---|---|---|
provider, model | empty | Use the route recorded for the step; configure both as a host fallback if needed. |
language | auto | Follow the DSH locale; overridable in the plugin settings. |
thinking | off | Disable thinking for the auxiliary request, falling back if unsupported. |
maxChars | 600 | Combined title/body character budget. |
maxOutputTokens | 220 | Auxiliary output token cap. |
minReasoningChars | 120 | Shorter sources skip generation. |
maxInputChars | 24000 | Keep the head and tail of longer inputs. |
timeoutMs | 45000 | Per-generation deadline. |
cacheSize | 400 | Host cache entries; 0 disables caching. |
Durable cache: ~/.dsh/thinking-summary/cache.json; plugin settings: ~/.dsh/thinking-summary/settings.json. Clearing the cache does not alter sessions.