reasoning-effort 34

Third-party provider reasoning-effort AND input-modality settings for DeepSeek Harness: thinking levels and image-input support declared per model, auto-adapted from a model knowledge base + wire-protocol inference, edited right inside the official Models

为 DSH 的 pi-ai 第三方模型补充可配置的思考强度档位，并设置子 agent 默认思考强度。

Per-model reasoning efforts and input modalities for DeepSeek Harness — one settings section, no YAML.

Provider-aware reasoning effort settings and composer control for DeepSeek Harness.

Model policy for DeepSeek Harness (DSH): set_model / list_models agent tools, preset mode with switch_preset / list_presets, a configurable stage-to-model system prompt, plan-mode auto-transition, provider whitelist and context-window guards, active-model

DeepSeek Harness plugin: reasoning-effort auto-fill declarations and progressive Figma MCP bridging (design-to-code + AI-driven design).

DSH plugin: fixes compaction failure on local qwen3.8-27b gateways — xhigh thinking burns the entire output token budget, so thinking is off for compaction-only, with the model's non-thinking sampling parameters

Capability-aware reasoning effort slider for the DeepSeek Harness web model selector.

Unified ModelID routing for DeepSeek Harness: one logical model id over multiple providers, first-token failover with cooldown, health-aware ranking, fallback reasoning efforts, purpose-based simple/complex task split, per-tier reasoning effort, and a set

DSH 的第三方模型补齐思考等级：在官方 Models 页每个模型行注入 Thinking 开关和强度/模态编辑器，在输入框模型菜单加供应商→模型选择器、视觉徽章与等级行，并带搜索、拖拽排序、供应商启用开关；以及按路由接管请求头与 user-agent。

DSH 自定义模型思考等级管理、渠道重试策略与偏好记忆插件：原生设计语言、全档位思考配置、渠道级重试控制与无感隔离式跨会话偏好记忆。

Provider-folded model selector for the DSH Web composer with catalog-driven reasoning effort choices.

DSH plugin: auto-detect and configure model capabilities (reasoningEfforts + capacities) for llm-pi-ai. Vision (input) is managed by the official UI. Successor to dsh-reasoning-efforts.

DeepSeek Harness 插件：为第三方模型提供商的模型自动匹配模型名称并补齐推理档位（reasoning effort），使思考强度选择器在官方模型选择器中可用；并在设置页提供「模型能力」页，可编辑每个已添加模型的上下文窗口/输出上限/图片多模态。Third-party DSH plugin that matches model ids against a built-in knowledge table, declares per-model reasoning tiers in the ll

Capability metadata for hand-declared gateway models, resolved at runtime by bare model name (context window, output cap, reasoning levels, vision) — matched against the official DeepSeek route's catalog, the installed pi-ai catalog, opencode and models.d

DeepSeek Harness 推理强度与网关增强插件：自动为 llm-pi-ai 下所有模型补齐推理强度档位（off→max 全七档），并自动为 OpenCode Go 网关注入必需的 x-opencode-session 路由 Header，写入 settings 由 dsh 原生解析生效。

DSH 模型整理插件：输入框的模型下拉改为「供应商分组 + 模型二级折叠菜单」（含推理等级），并在模型设置页提供一个可拖拽的「供应商与模型顺序」浮窗。

DSH (DeepSeek Harness) 插件：为每个 LLM 模型配置默认推理等级（reasoningEffort），并在请求进入底层调用链前按路由规则自动补齐

T3 Code's New Session screen for DeepSeek Harness: the "What should we build in <project>?" headline with an inline project dropdown, a reasoning-effort picker and a git branch picker in the composer.

Model Ultra — DSH plugin for llm-pi-ai provider lifecycle: create/edit, custom headers, enable/disable with uninstall-restore safety, per-model thinking effort, OpenRouter provider routing, smart routing, composites, observability, and per-model connectiv

DeepSeek Harness plugin: subagents with their own model and reasoning effort 闂?GUI settings card (plugin-owned config file), config defaults, and a parent-model-controlled delegation tool (subagent_select).

Codex-style reasoning-effort slider for the DeepSeek Harness composer: pick a model, then drag its thinking level on a gradient particle rail.

Configure model input capabilities and reasoning efforts for DeepSeek Harness

DeepSeek Harness plugin to audit and correct llm-pi-ai model capability declarations by probing the endpoint itself: out-of-range maxTokens, contextWindow, reasoning levels, image input.

Per-delegation reasoning effort for dsh subagents: makes AgentOptions.reasoningEffort reach child requests (the agent loop silently ignored it) and adds a subagent_effort tool that accepts reasoning_effort and model per call

Codex-style DeepSeek Harness model and reasoning selector with model-advertised effort levels, DSH-native themes, left-clipped radiation effects, and copy-ready reasoning-effort declaration guidance for custom-provider models.

DSH 模型整理插件：输入框的模型下拉改为「供应商分组 + 模型二级折叠菜单」（含推理等级），并在模型设置页提供一个可拖拽的「供应商与模型顺序」浮窗。

dsh web 插件：自动配置模型思考档位与能力，并按 provider+model 记住最后选择的思考档位

DeepSeek Harness model selector upgrade: one-level model & reasoning-effort picker with 5 Liang tiers, custom model aliases, and peak/off-peak price countdown.

DSH plugin (NInfer engine only): fixes compaction failure on local qwen3.8-27b gateways served by NInfer — xhigh thinking burns the entire output token budget, so thinking is off for compaction-only, with the model's non-thinking sampling parameters; the

Model router for DeepSeek Harness: keep a pool of models and every step is routed to one of them with its reasoning effort set — a click-only composer panel, three authorization scopes (a manual pick only counts in the scope that owns it, and effort scope

Auto-adjust model reasoning effort per task: off/low/high/max with task classification, peak-pricing-period capping, countdown notices, and in-turn error escalation

DSH plugin: read reasoning_efforts from OpenAI-compatible (vLLM) providers' /models listing and sync them into the llm-pi-ai settings section so the Web composer shows a per-model reasoning-effort picker.

Third-party provider reasoning-effort AND input-modality settings for DeepSeek Harness: thinking levels and image-input support declared per model, auto-adapted from a model knowledge base + wire-protocol inference, edited right inside the official Models