oh-my-opencode-slim 2.1.0 → 2.2.0

package/README.md

@@ -30,19 +30,14 @@ The main idea is simple: instead of forcing one model to do everything, the plug
 
 To explore the agents themselves, see **[Meet the Pantheon](#meet-the-pantheon)**. For the full feature set, see **[Features & Workflows](#features-and-workflows)** below.
 
-### Manage Agent Skills with LazySkills
+### OpenAI GPT-5.6
 
 <p align="center">
-  <a href="https://github.com/alvinunreal/lazyskills">
-    <img src="img/lazyskills-wide.svg" alt="LazySkills" width="720">
-  </a>
+  <img src="img/openai-gpt-5-6-pantheon.jpeg" alt="OpenAI GPT-5.6 Pantheon: Terra, Sol, and Luna" width="100%">
 </p>
 
-**[LazySkills](https://github.com/alvinunreal/lazyskills)** is a terminal UI for managing agent skills. It gives you one place to see what is installed, which agents can use each skill, why visibility may be broken, and what actions are safe to run next.
-
-<p align="center">
-  <a href="https://github.com/alvinunreal/lazyskills"><b>Explore LazySkills →</b></a>
-</p>
+The default [OpenAI preset](docs/openai-preset.md) maps Terra to Orchestrator,
+Sol to Oracle, and Luna to the fast specialist lanes.
 
 ### What Users Say
 
@@ -137,7 +132,7 @@ Then:
 > It's **recommended** to understand how background orchestration works. The **[Orchestrator prompt](https://github.com/alvinunreal/oh-my-opencode-slim/blob/master/src/agents/orchestrator.ts#L28)** contains the scheduler rules, specialist routing logic, and thresholds for when work should be assigned to background agents. You can always delegate manually by calling a subagent via: `@agentName <task>`
 
 > [!TIP]
-> Because background agents are now the default workflow, it is **highly recommended** to enable and configure **[Multiplexer Integration](docs/multiplexer-integration.md)**. It automatically opens each agent in a dedicated Tmux, Zellij, or Herdr pane, so you can watch specialists work live while the Orchestrator continues coordinating the session.
+> Because background agents are now the default workflow, it is **highly recommended** to enable and configure **[Multiplexer Integration](docs/multiplexer-integration.md)**. It automatically opens each agent in a dedicated Tmux, Zellij, Herdr, or cmux pane, so you can watch specialists work live while the Orchestrator continues coordinating the session.
 
 The default generated configuration includes both `openai` and `opencode-go` presets.
 
@@ -147,12 +142,12 @@ The default generated configuration includes both `openai` and `opencode-go` pre
   "preset": "openai",
   "presets": {
     "openai": {
-      "orchestrator": { "model": "openai/gpt-5.5", "variant": "medium", "skills": ["*"], "mcps": ["*", "!context7"] },
-      "oracle": { "model": "openai/gpt-5.5", "variant": "high", "skills": ["simplify"], "mcps": [] },
-      "librarian": { "model": "openai/gpt-5.4-mini", "variant": "low", "skills": [], "mcps": ["websearch", "context7", "gh_grep"] },
-      "explorer": { "model": "openai/gpt-5.4-mini", "variant": "low", "skills": [], "mcps": [] },
-      "designer": { "model": "openai/gpt-5.4-mini", "variant": "medium", "skills": [], "mcps": [] },
-      "fixer": { "model": "openai/gpt-5.5", "variant": "low", "skills": [], "mcps": [] }
+      "orchestrator": { "model": "openai/gpt-5.6-terra", "variant": "medium", "skills": ["*"], "mcps": ["*", "!context7"] },
+      "oracle": { "model": "openai/gpt-5.6-sol", "variant": "high", "skills": ["simplify"], "mcps": [] },
+      "librarian": { "model": "openai/gpt-5.6-luna", "variant": "low", "skills": [], "mcps": ["websearch", "context7", "gh_grep"] },
+      "explorer": { "model": "openai/gpt-5.6-luna", "variant": "low", "skills": [], "mcps": [] },
+      "designer": { "model": "openai/gpt-5.6-luna", "variant": "medium", "skills": [], "mcps": [] },
+      "fixer": { "model": "openai/gpt-5.6-luna", "variant": "medium", "skills": [], "mcps": [] }
     },
     "opencode-go": {
       "orchestrator": { "model": "opencode-go/glm-5.2", "skills": [ "*" ], "mcps": [ "*", "!context7" ] },
@@ -166,9 +161,17 @@ The default generated configuration includes both `openai` and `opencode-go` pre
 }
 ```
 
+### Preset Docs
+
+- **[OpenAI Preset](docs/openai-preset.md)** — the default generated preset; runs all agents on OpenAI models.
+- **[OpenCode Go Preset](docs/opencode-go-preset.md)** — runs the agents on OpenCode Go models; enables the Observer agent for visual analysis since its orchestrator model isn't multimodal.
+- **[Author's Preset](docs/authors-preset.md)** — the exact config the author runs day to day, with third-party skills.
+- **[$30 Preset](docs/thirty-dollars-preset.md)** — a mixed-provider setup built around Codex Plus and GitHub Copilot Pro for about $30/month.
+- **[OpenCode Zen Free Preset](docs/opencode-zen-free-preset.md)** — every agent runs on an opencode free model; no usage cost.
+
 ### For Alternative Providers
 
-To use custom providers or a mixed-provider setup, use **[Configuration](docs/configuration.md)** for the full reference. If you want a ready-made starting point, check the **[Author's Preset](docs/authors-preset.md)** and **[$30 Preset](docs/thirty-dollars-preset.md)** - the `$30` preset is the best cheap setup.
+To use custom providers or a mixed-provider setup, use **[Configuration](docs/configuration.md)** for the full reference.
 
 ### ✅ Verify Your Setup
 
@@ -206,6 +209,9 @@ verification while specialists do the work in their own lanes.
   agents are currently active, including parallel background specialists.
 - **[Deepwork](#deepwork)** - a structured workflow for large, multi-file, risky,
   or phased coding work using persistent plan files and Oracle review gates.
+- **[Verification Planning](#verification-planning)** - plans a project-specific
+  evidence path before non-trivial implementation, including verification
+  affordances when the system needs to become more legible to an agent.
 - **[Reflect](#reflect)** - reviews repeated work patterns and suggests reusable skills,
   agents, commands, config rules, prompt rules, or project playbooks.
 - **[Worktrees](#worktrees)** - manages Git worktrees as isolated coding lanes
@@ -220,7 +226,7 @@ V2 makes background specialists the default mental model: the Orchestrator plans
 the work graph, launches the right agents, avoids overlapping write ownership,
 and waits for terminal task results before acting on them.
 
-See **[Background Orchestration](docs/v2-background-orchestration.md)** for the
+See **[Background Orchestration](docs/background-orchestration.md)** for the
 full scheduler model.
 
 #### Companion
@@ -260,6 +266,18 @@ Start it with:
 See **[Skills](docs/skills.md#deepwork)** for when to use it and how the workflow
 runs.
 
+#### Verification Planning
+
+Verification Planning gives the Orchestrator a way to decide how a non-trivial
+change will be proven before implementation begins. It frames the claim, designs
+project-specific evidence paths, and can create a small verification affordance
+when the system cannot directly reveal the decisive state to an agent. For
+unfamiliar capabilities, it asks the Librarian for focused research before
+choosing an approach.
+
+See **[Skills](docs/skills.md#verification-planning)** for the evidence-path
+workflow and its safety boundaries.
+
 #### Reflect
 
 Reflect helps the Orchestrator learn from repeated workflow friction. It reviews
@@ -335,12 +353,12 @@ rules.
   </tr>
   <tr>
     <td colspan="2">
-      <b>Default Model:</b> <code>openai/gpt-5.5 (medium)</code>
+      <b>Default Model:</b> <code>openai/gpt-5.6-terra (medium)</code>
     </td>
   </tr>
   <tr>
     <td colspan="2">
-      <b>Recommended Models:</b> <code>openai/gpt-5.5 (medium)</code> <code>anthropic/claude-fable-5</code> <code>anthropic/claude-opus-4-8</code>
+      <b>Recommended Models:</b> <code>openai/gpt-5.6-terra (medium)</code> <code>anthropic/claude-fable-5</code> <code>anthropic/claude-opus-4-8</code>
     </td>
   </tr>
   <tr>
@@ -376,7 +394,7 @@ rules.
   </tr>
   <tr>
     <td colspan="2">
-      <b>Default Model:</b> <code>openai/gpt-5.4-mini</code>
+      <b>Default Model:</b> <code>openai/gpt-5.6-luna</code>
     </td>
   </tr>
   <tr>
@@ -417,12 +435,12 @@ rules.
   </tr>
   <tr>
     <td colspan="2">
-      <b>Default Model:</b> <code>openai/gpt-5.5 (high)</code>
+      <b>Default Model:</b> <code>openai/gpt-5.6-sol (high)</code>
     </td>
   </tr>
   <tr>
     <td colspan="2">
-      <b>Recommended Models:</b> <code>openai/gpt-5.5 (xhigh)</code> <code>anthropic/claude-fable-5</code> <code>anthropic/claude-opus-4-8 (xhigh)</code>
+      <b>Recommended Models:</b> <code>openai/gpt-5.6-sol (xhigh)</code> <code>anthropic/claude-fable-5</code> <code>anthropic/claude-opus-4-8 (xhigh)</code>
     </td>
   </tr>
   <tr>
@@ -507,7 +525,7 @@ rules.
   </tr>
   <tr>
     <td colspan="2">
-      <b>Default Model:</b> <code>openai/gpt-5.4-mini</code>
+      <b>Default Model:</b> <code>openai/gpt-5.6-luna</code>
     </td>
   </tr>
   <tr>
@@ -548,7 +566,7 @@ rules.
   </tr>
   <tr>
     <td colspan="2">
-      <b>Default Model:</b> <code>openai/gpt-5.4-mini</code>
+      <b>Default Model:</b> <code>openai/gpt-5.6-luna</code>
     </td>
   </tr>
   <tr>
@@ -589,12 +607,12 @@ rules.
   </tr>
   <tr>
     <td colspan="2">
-      <b>Default Model:</b> <code>openai/gpt-5.5 (low)</code>
+      <b>Default Model:</b> <code>openai/gpt-5.6-luna (medium)</code>
     </td>
   </tr>
   <tr>
     <td colspan="2">
-      <b>Recommended Models:</b> <code>openai/gpt-5.5 (low)</code> <code>anthropic/claude-sonnet-4-6</code>
+      <b>Recommended Models:</b> <code>openai/gpt-5.6-luna (medium)</code> <code>anthropic/claude-sonnet-4-6</code>
     </td>
   </tr>
   <tr>
@@ -611,7 +629,7 @@ rules.
 ### Observer: The Silent Witness
 
 > [!NOTE]
-> **Why a separate agent?** If your Orchestrator model is not multimodal, enable Observer to handle images, screenshots, PDFs, and other visual files. Observer is disabled by default and gives the Orchestrator a dedicated multimodal reader without forcing you to change your main reasoning model. Set `disabled_agents: []` and an `observer` model in your configuration. The bundled `opencode-go` install preset does this automatically because its GLM Orchestrator is not multimodal.
+> **Why a separate agent?** If your Orchestrator model is not multimodal, enable Observer to handle images, screenshots, and other visual files. Observer is disabled by default and gives the Orchestrator a dedicated multimodal reader without forcing you to change your main reasoning model. Set `disabled_agents: []` and an `observer` model in your configuration. The bundled `opencode-go` install preset does this automatically because its GLM Orchestrator is not multimodal. Omitting `image_routing` preserves existing conditional Observer behavior. Set `image_routing: "auto"` only when Observer is enabled, or `"direct"` to always pass image attachments to the Orchestrator.
 
 <table>
   <tr>
@@ -625,7 +643,7 @@ rules.
 
 - Images, screenshots, diagrams → `read` tool (native image support)
 - PDFs and binary documents → `read` tool (text + structure extraction)
-- **Disabled by default** - enable with `"disabled_agents": []` and configure a vision-capable model; installing with `--preset=opencode-go` enables it with `opencode-go/kimi-k2.6`
+- **Disabled by default** - enable with `"disabled_agents": []` and configure a vision-capable model; installing with `--preset=opencode-go` enables it with `opencode-go/kimi-k2.6`. Image attachments route to Observer by default when it is enabled; set `"image_routing": "direct"` to keep them on the Orchestrator.
 
     </td>
   </tr>
@@ -636,7 +654,7 @@ rules.
   </tr>
   <tr>
     <td colspan="2">
-      <b>Default Model:</b> <code>openai/gpt-5.4-mini</code> - <i>configure a vision-capable model to enable</i>
+      <b>Default Model:</b> <code>openai/gpt-5.6-luna</code> - <i>configure a vision-capable model to enable</i>
     </td>
   </tr>
   <tr>
@@ -661,7 +679,7 @@ Use this section as a map: start with installation, then jump to features, confi
 | **[Council](docs/council.md)** | Run multiple models in parallel and synthesize a single answer with `@council` |
 | **[Custom Agents](docs/configuration.md#custom-agents)** | Define your own specialists with custom prompts, models, MCP access, and Orchestrator delegation rules |
 | **[ACP Agents](docs/acp-agents.md)** | Connect external ACP-compatible agents such as Claude Code ACP or Gemini ACP as delegatable subagents |
-| **[Multiplexer Integration](docs/multiplexer-integration.md)** | Watch agents work live in Tmux, Zellij, or Herdr panes |
+| **[Multiplexer Integration](docs/multiplexer-integration.md)** | Watch agents work live in Tmux, Zellij, Herdr, or cmux panes |
 | **[Codemap](docs/codemap.md)** | Generate hierarchical codemaps to understand large codebases faster |
 | **[Clonedeps](docs/clonedeps.md)** | Clone selected dependency source into an ignored local workspace for inspection |
 | **[Worktrees](docs/worktrees.md)** | Use `.slim/worktrees/` lanes for isolated parallel or risky coding work |
@@ -678,7 +696,7 @@ Use this section as a map: start with installation, then jump to features, confi
 | **[Project Customization](docs/project-local-customization.md)** | Repository-specific custom agents, prompt overrides, per-agent skills, and precedence |
 | **[Background Orchestration](docs/background-orchestration.md)** | Scheduler-first orchestrator model built around native background subagents |
 | **[Maintainer Guide](docs/maintainers.md)** | Issue triage rules, label meanings, support routing, and repo maintenance workflow |
-| **[Skills](docs/skills.md)** | Bundled skills such as `simplify`, `codemap`, `clonedeps`, `deepwork`, `reflect`, `worktrees`, and `oh-my-opencode-slim` |
+| **[Skills](docs/skills.md)** | Bundled skills such as `simplify`, `codemap`, `clonedeps`, `deepwork`, `verification-planning`, `reflect`, `worktrees`, and `oh-my-opencode-slim` |
 | **[MCPs](docs/mcps.md)** | `websearch`, `context7`, `gh_grep`, and how MCP permissions work per agent |
 | **[Tools](docs/tools.md)** | Built-in tool capabilities like `webfetch`, LSP tools, code search, and formatters |
 
@@ -699,7 +717,7 @@ Use this section as a map: start with installation, then jump to features, confi
   <p><sub>Every merged contribution leaves a mark on the realm.</sub></p>
 
   <!-- ALL-CONTRIBUTORS-BADGE:START - Do not remove or modify this section -->
-[![All Contributors](https://img.shields.io/badge/all_contributors-72-orange.svg?style=flat-square)](#contributors-)
+[![All Contributors](https://img.shields.io/badge/all_contributors-76-orange.svg?style=flat-square)](#contributors-)
 <!-- ALL-CONTRIBUTORS-BADGE:END -->
 </div>
 
@@ -806,6 +824,12 @@ Use this section as a map: start with installation, then jump to features, confi
       <td align="center" valign="top" width="16.66%"><a href="https://rgutzen.github.io/"><img src="https://avatars.githubusercontent.com/u/16289604?v=4?s=100" width="100px;" alt="Robin Gutzen"/><br /><sub><b>Robin Gutzen</b></sub></a><br /><a href="https://github.com/alvinunreal/oh-my-opencode-slim/commits?author=rgutzen" title="Code">💻</a></td>
       <td align="center" valign="top" width="16.66%"><a href="https://github.com/dragon-Elec"><img src="https://avatars.githubusercontent.com/u/197374270?v=4?s=100" width="100px;" alt="Yash"/><br /><sub><b>Yash</b></sub></a><br /><a href="https://github.com/alvinunreal/oh-my-opencode-slim/commits?author=dragon-Elec" title="Code">💻</a></td>
     </tr>
+    <tr>
+      <td align="center" valign="top" width="16.66%"><a href="https://github.com/Jiajun0413"><img src="https://avatars.githubusercontent.com/u/184531967?v=4?s=100" width="100px;" alt="Liu Jiajun"/><br /><sub><b>Liu Jiajun</b></sub></a><br /><a href="https://github.com/alvinunreal/oh-my-opencode-slim/commits?author=Jiajun0413" title="Code">💻</a></td>
+      <td align="center" valign="top" width="16.66%"><a href="https://github.com/umi008"><img src="https://avatars.githubusercontent.com/u/200843810?v=4?s=100" width="100px;" alt="Ulises Millán"/><br /><sub><b>Ulises Millán</b></sub></a><br /><a href="https://github.com/alvinunreal/oh-my-opencode-slim/commits?author=umi008" title="Code">💻</a></td>
+      <td align="center" valign="top" width="16.66%"><a href="https://github.com/HighColdHC"><img src="https://avatars.githubusercontent.com/u/35870222?v=4?s=100" width="100px;" alt="HighColdHC"/><br /><sub><b>HighColdHC</b></sub></a><br /><a href="https://github.com/alvinunreal/oh-my-opencode-slim/commits?author=HighColdHC" title="Code">💻</a></td>
+      <td align="center" valign="top" width="16.66%"><a href="https://hardcore.engineer/about"><img src="https://avatars.githubusercontent.com/u/401815?v=4?s=100" width="100px;" alt="Stephan Schielke"/><br /><sub><b>Stephan Schielke</b></sub></a><br /><a href="https://github.com/alvinunreal/oh-my-opencode-slim/commits?author=stephanschielke" title="Code">💻</a></td>
+    </tr>
   </tbody>
 </table>
 

package/README.zh-CN.md

@@ -32,19 +32,13 @@ oh-my-opencode-slim 是一个用于 OpenCode 的智能体编排插件。它内
 
 要了解智能体本身，请参阅 **[认识众神殿](#meet-the-pantheon)**。如需了解完整特性集，请参阅下方的 **[特性与工作流](#features-and-workflows)**。
 
-### 用 LazySkills 管理智能体技能
+### OpenAI GPT-5.6
 
 <p align="center">
-  <a href="https://github.com/alvinunreal/lazyskills">
-    <img src="img/lazyskills-wide.svg" alt="LazySkills" width="720">
-  </a>
+  <img src="img/openai-gpt-5-6-pantheon.jpeg" alt="OpenAI GPT-5.6 众神殿：Terra、Sol 和 Luna" width="720">
 </p>
 
-**[LazySkills](https://github.com/alvinunreal/lazyskills)** 是一个用于管理智能体技能的终端 UI。它让你可以在一个地方查看已安装的技能、哪些智能体可以使用每个技能、为什么可见性可能失效，以及接下来可以安全执行哪些操作。
-
-<p align="center">
-  <a href="https://github.com/alvinunreal/lazyskills"><b>探索 LazySkills →</b></a>
-</p>
+默认的 [OpenAI 预设](docs/openai-preset.md) 将 Terra 映射为 Orchestrator、Sol 映射为 Oracle、Luna 映射为快速专家通道。
 
 ### 用户怎么说
 
@@ -134,7 +128,7 @@ bun run build
 > **建议**了解后台编排的工作原理。**[编排者提示词 (Orchestrator prompt)](https://github.com/alvinunreal/oh-my-opencode-slim/blob/master/src/agents/orchestrator.ts#L28)** 包含调度规则、专家路由逻辑，以及何时应把工作分配给后台智能体的阈值。您始终可以通过以下方式手动委派任务：`@智能体名称 <任务内容>`
 
 > [!TIP]
-> 由于后台智能体现在是默认工作流，**强烈建议**启用并配置 **[Multiplexer Integration](docs/multiplexer-integration.md)**。它会自动在专用的 Tmux、Zellij 或 Herdr 窗格中打开每个智能体，让您在 Orchestrator 继续协调会话时，实时跟进各个专家智能体的工作。
+> 由于后台智能体现在是默认工作流，**强烈建议**启用并配置 **[Multiplexer Integration](docs/multiplexer-integration.md)**。它会自动在专用的 Tmux、Zellij、Herdr 或 cmux 窗格中打开每个智能体，让您在 Orchestrator 继续协调会话时，实时跟进各个专家智能体的工作。
 
 默认生成的配置包含 `openai` 和 `opencode-go` 两个预设：
 
@@ -144,12 +138,12 @@ bun run build
   "preset": "openai",
   "presets": {
     "openai": {
-      "orchestrator": { "model": "openai/gpt-5.5", "variant": "medium", "skills": ["*"], "mcps": ["*", "!context7"] },
-      "oracle": { "model": "openai/gpt-5.5", "variant": "high", "skills": ["simplify"], "mcps": [] },
-      "librarian": { "model": "openai/gpt-5.4-mini", "variant": "low", "skills": [], "mcps": ["websearch", "context7", "gh_grep"] },
-      "explorer": { "model": "openai/gpt-5.4-mini", "variant": "low", "skills": [], "mcps": [] },
-      "designer": { "model": "openai/gpt-5.4-mini", "variant": "medium", "skills": [], "mcps": [] },
-      "fixer": { "model": "openai/gpt-5.5", "variant": "low", "skills": [], "mcps": [] }
+      "orchestrator": { "model": "openai/gpt-5.6-terra", "variant": "medium", "skills": ["*"], "mcps": ["*", "!context7"] },
+      "oracle": { "model": "openai/gpt-5.6-sol", "variant": "high", "skills": ["simplify"], "mcps": [] },
+      "librarian": { "model": "openai/gpt-5.6-luna", "variant": "low", "skills": [], "mcps": ["websearch", "context7", "gh_grep"] },
+      "explorer": { "model": "openai/gpt-5.6-luna", "variant": "low", "skills": [], "mcps": [] },
+      "designer": { "model": "openai/gpt-5.6-luna", "variant": "medium", "skills": [], "mcps": [] },
+      "fixer": { "model": "openai/gpt-5.6-luna", "variant": "medium", "skills": [], "mcps": [] }
     },
     "opencode-go": {
       "orchestrator": { "model": "opencode-go/glm-5.2", "skills": [ "*" ], "mcps": [ "*", "!context7" ] },
@@ -197,6 +191,7 @@ V2 将 oh-my-opencode-slim 变成了以调度器为核心的多智能体工作
 - **[后台智能体](#后台智能体)** - Orchestrator 现在会把专家作为后台任务派发，跟踪任务/会话 ID，等待完成事件，并在继续之前整合结果。
 - **[Companion](#companion)** - 可选的浮动桌面窗口会显示当前活跃的智能体，包括并行运行的后台专家。
 - **[Deepwork](#deepwork)** - 面向大型、多文件、高风险或分阶段编码工作的结构化工作流，使用持久化计划文件和 Oracle 评审关卡。
+- **[验证规划](#验证规划)** - 在非平凡实现前规划项目特定的证据路径；当系统需要更易被智能体理解时，可加入验证能力。
 - **[Reflect](#reflect)** - 回顾重复出现的工作模式，并建议可复用的 skill、智能体、命令、配置规则、提示词规则或项目 playbook。
 - **[Worktrees](#worktrees)** - 将 Git worktree 作为隔离编码通道管理，并为复杂、高风险或并行任务提供安全协议。
 - **[oh-my-opencode-slim Skill](#oh-my-opencode-slim-skill)** - 随包提供的配置技能，可安全调优模型、提示词、自定义智能体、MCP 访问、预设和插件行为。
@@ -236,6 +231,12 @@ Deepwork 适用于重型编码会话：大范围重构、多阶段功能、高
 
 何时使用以及工作流如何运行，请参阅 **[Skills](docs/skills.md#deepwork)**。
 
+#### 验证规划
+
+验证规划让 Orchestrator 在开始非平凡实现前，先决定如何证明变更有效。它会界定要成立的主张、设计项目特定的证据路径；当系统无法直接向智能体揭示决定性状态时，还可创建小型验证能力。面对不熟悉的能力时，它会在选择方案前请 Librarian 做有针对性的研究。
+
+证据路径工作流及其安全边界见 **[Skills](docs/skills.md#verification-planning)**。
+
 #### Reflect
 
 Reflect 帮助 Orchestrator 从重复出现的工作流摩擦中学习。它会回顾近期工作和现有资产，然后建议最小且有用的改进：skill、自定义智能体、命令、配置规则、提示词规则、MCP 权限变更或项目 playbook。如果证据不足，它应建议什么都不创建。
@@ -303,12 +304,12 @@ Worktrees 将 Git worktree 作为安全、隔离的编码通道管理，默认
   </tr>
   <tr>
     <td colspan="2">
-      <b>默认模型：</b> <code>openai/gpt-5.5 (medium)</code>
+      <b>默认模型：</b> <code>openai/gpt-5.6-terra (medium)</code>
     </td>
   </tr>
   <tr>
     <td colspan="2">
-      <b>推荐模型：</b> <code>openai/gpt-5.5 (medium)</code> <code>anthropic/claude-opus-4.6</code>
+      <b>推荐模型：</b> <code>openai/gpt-5.6-terra (medium)</code> <code>anthropic/claude-opus-4.6</code>
     </td>
   </tr>
   <tr>
@@ -344,7 +345,7 @@ Worktrees 将 Git worktree 作为安全、隔离的编码通道管理，默认
   </tr>
   <tr>
     <td colspan="2">
-      <b>默认模型：</b> <code>openai/gpt-5.4-mini</code>
+      <b>默认模型：</b> <code>openai/gpt-5.6-luna</code>
     </td>
   </tr>
   <tr>
@@ -385,12 +386,12 @@ Worktrees 将 Git worktree 作为安全、隔离的编码通道管理，默认
   </tr>
   <tr>
     <td colspan="2">
-      <b>默认模型：</b> <code>openai/gpt-5.5 (high)</code>
+      <b>默认模型：</b> <code>openai/gpt-5.6-sol (high)</code>
     </td>
   </tr>
   <tr>
     <td colspan="2">
-      <b>推荐模型：</b> <code>openai/gpt-5.5 (high)</code> <code>google/gemini-3.1-pro-preview (high)</code>
+      <b>推荐模型：</b> <code>openai/gpt-5.6-sol (high)</code> <code>google/gemini-3.1-pro-preview (high)</code>
     </td>
   </tr>
   <tr>
@@ -475,7 +476,7 @@ Worktrees 将 Git worktree 作为安全、隔离的编码通道管理，默认
   </tr>
   <tr>
     <td colspan="2">
-      <b>默认模型：</b> <code>openai/gpt-5.4-mini</code>
+      <b>默认模型：</b> <code>openai/gpt-5.6-luna</code>
     </td>
   </tr>
   <tr>
@@ -516,7 +517,7 @@ Worktrees 将 Git worktree 作为安全、隔离的编码通道管理，默认
   </tr>
   <tr>
     <td colspan="2">
-      <b>默认模型：</b> <code>openai/gpt-5.4-mini</code>
+      <b>默认模型：</b> <code>openai/gpt-5.6-luna</code>
     </td>
   </tr>
   <tr>
@@ -557,12 +558,12 @@ Worktrees 将 Git worktree 作为安全、隔离的编码通道管理，默认
   </tr>
   <tr>
     <td colspan="2">
-      <b>默认模型：</b> <code>openai/gpt-5.5 (low)</code>
+      <b>默认模型：</b> <code>openai/gpt-5.6-luna (medium)</code>
     </td>
   </tr>
   <tr>
     <td colspan="2">
-      <b>推荐模型：</b> <code>openai/gpt-5.5 (low)</code>
+      <b>推荐模型：</b> <code>openai/gpt-5.6-luna (medium)</code>
     </td>
   </tr>
   <tr>
@@ -604,7 +605,7 @@ Worktrees 将 Git worktree 作为安全、隔离的编码通道管理，默认
   </tr>
   <tr>
     <td colspan="2">
-      <b>默认模型：</b> <code>openai/gpt-5.4-mini</code> - <i>需配置具有视觉能力的模型以启用</i>
+      <b>默认模型：</b> <code>openai/gpt-5.6-luna</code> - <i>需配置具有视觉能力的模型以启用</i>
     </td>
   </tr>
   <tr>
@@ -629,7 +630,7 @@ Worktrees 将 Git worktree 作为安全、隔离的编码通道管理，默认
 | **[Council](docs/council.md)** | 使用 `@council` 并行运行多个模型并合成单一答案 |
 | **[自定义智能体](docs/configuration.md#custom-agents)** | 使用自定义提示词、模型、MCP 访问和 Orchestrator 委派规则定义自己的专家 |
 | **[ACP Agents](docs/acp-agents.md)** | 将 Claude Code ACP 或 Gemini ACP 等外部 ACP 兼容智能体连接为可委派子智能体 |
-| **[多路复用器集成](docs/multiplexer-integration.md)** | 在 Tmux、Zellij 或 Herdr 窗格中实时观看智能体工作 |
+| **[多路复用器集成](docs/multiplexer-integration.md)** | 在 Tmux、Zellij、Herdr 或 cmux 窗格中实时观看智能体工作 |
 | **[Codemap](docs/codemap.md)** | 生成层级代码地图，更快理解大型代码库 |
 | **[Clonedeps](docs/clonedeps.md)** | 将选定的依赖源码克隆到被忽略的本地工作区中以供检查 |
 | **[Worktrees](docs/worktrees.md)** | 使用 `.slim/worktrees/` 通道进行隔离的并行或高风险编码工作 |
@@ -645,7 +646,7 @@ Worktrees 将 Git worktree 作为安全、隔离的编码通道管理，默认
 | **[配置](docs/configuration.md)** | 配置文件位置、JSONC 支持、提示词覆盖和完整选项参考 |
 | **[后台编排](docs/background-orchestration.md)** | 围绕原生后台子智能体构建的调度器优先 Orchestrator 模型 |
 | **[维护者指南](docs/maintainers.md)** | issue 分流规则、标签含义、支持路由和仓库维护工作流 |
-| **[Skills](docs/skills.md)** | `simplify`、`codemap`、`clonedeps`、`deepwork`、`reflect`、`worktrees` 和 `oh-my-opencode-slim` 等捆绑技能 |
+| **[Skills](docs/skills.md)** | `simplify`、`codemap`、`clonedeps`、`deepwork`、`verification-planning`、`reflect`、`worktrees` 和 `oh-my-opencode-slim` 等捆绑技能 |
 | **[MCPs](docs/mcps.md)** | `websearch`、`context7`、`gh_grep` 以及每个智能体的 MCP 权限机制 |
 | **[Tools](docs/tools.md)** | `webfetch`、LSP 工具、代码搜索和格式化工具等内置工具能力 |
 

package/dist/agents/index.d.ts

@@ -29,7 +29,3 @@ export declare function getAgentConfigs(config?: PluginConfig, options?: {
  * Get the set of disabled agent names from config, applying protection rules.
  */
 export declare function getDisabledAgents(config?: PluginConfig): Set<string>;
-/**
- * Get the list of enabled (non-disabled) agent names.
- */
-export declare function getEnabledAgentNames(config?: PluginConfig): string[];

package/dist/cli/index.js

@@ -293,14 +293,43 @@ var ALL_AGENT_NAMES = ["orchestrator", ...SUBAGENT_NAMES];
 var PROTECTED_AGENTS = new Set(["orchestrator", "councillor"]);
 var MAX_POLL_TIME_MS = 5 * 60 * 1000;
 var PHASE_REMINDER_TEXT = `!IMPORTANT! Scheduler workflow: plan lanes/dependencies → dispatch background specialists → track task IDs → wait for hook-driven completion → reconcile terminal results → verify. Do not poll running jobs, consume running-job output, or advance dependent work. !END!`;
-var PHASE_REMINDER = `<internal_reminder>${PHASE_REMINDER_TEXT}</internal_reminder>`;
+function formatSystemReminder(text) {
+  return `<system-reminder>
+${text}
+</system-reminder>`;
+}
+var PHASE_REMINDER = formatSystemReminder(PHASE_REMINDER_TEXT);
 // src/config/council-schema.ts
 import { z } from "zod";
-var ModelIdSchema = z.string().regex(/^[^/\s]+\/[^\s]+$/, 'Expected provider/model format (e.g. "openai/gpt-5.4-mini")');
+
+// src/utils/councillor-models.ts
+function normalizeCouncillorModels(model, fallbackVariant) {
+  const raw = Array.isArray(model) ? model : [model];
+  return raw.map((entry) => typeof entry === "string" ? { id: entry, variant: fallbackVariant } : { id: entry.id, variant: entry.variant ?? fallbackVariant });
+}
+
+// src/config/council-schema.ts
+var ModelIdSchema = z.string().regex(/^[^/\s]+\/[^\s]+$/, 'Expected provider/model format (e.g. "openai/gpt-5.6-luna")');
+var CouncillorModelEntrySchema = z.object({
+  id: ModelIdSchema,
+  variant: z.string().optional()
+});
+var CouncillorModelSchema = z.union([
+  ModelIdSchema,
+  z.array(z.union([ModelIdSchema, CouncillorModelEntrySchema])).min(1)
+]).describe('Model ID in provider/model format (e.g. "openai/gpt-5.6-luna"), or an ' + "ordered fallback chain (array of model IDs or { id, variant } entries) " + "tried in order until one responds.");
 var CouncillorConfigSchema = z.object({
-  model: ModelIdSchema.describe('Model ID in provider/model format (e.g. "openai/gpt-5.4-mini")'),
+  model: CouncillorModelSchema,
   variant: z.string().optional(),
   prompt: z.string().optional().describe("Optional role/guidance injected into the councillor user prompt")
+}).transform((c) => {
+  const models = normalizeCouncillorModels(c.model, c.variant);
+  return {
+    model: models[0].id,
+    variant: c.variant,
+    prompt: c.prompt,
+    models
+  };
 });
 var CouncilPresetSchema = z.record(z.string(), z.record(z.string(), z.unknown())).transform((entries, ctx) => {
   const councillors = {};
@@ -386,6 +415,33 @@ var ManualPlanSchema = z2.object({
   librarian: ManualAgentPlanSchema,
   fixer: ManualAgentPlanSchema
 }).strict();
+var PermissionActionSchema = z2.enum(["ask", "allow", "deny"]);
+var PermissionRuleSchema = z2.union([
+  PermissionActionSchema,
+  z2.record(z2.string(), PermissionActionSchema)
+]);
+var PermissionObjectSchema = z2.object({
+  read: PermissionRuleSchema.optional(),
+  edit: PermissionRuleSchema.optional(),
+  glob: PermissionRuleSchema.optional(),
+  grep: PermissionRuleSchema.optional(),
+  list: PermissionRuleSchema.optional(),
+  bash: PermissionRuleSchema.optional(),
+  task: PermissionRuleSchema.optional(),
+  external_directory: PermissionRuleSchema.optional(),
+  lsp: PermissionRuleSchema.optional(),
+  skill: PermissionRuleSchema.optional(),
+  todowrite: PermissionActionSchema.optional(),
+  question: PermissionActionSchema.optional(),
+  webfetch: PermissionActionSchema.optional(),
+  websearch: PermissionActionSchema.optional(),
+  codesearch: PermissionActionSchema.optional(),
+  doom_loop: PermissionActionSchema.optional()
+}).catchall(PermissionRuleSchema);
+var PermissionConfigSchema = z2.union([
+  PermissionActionSchema,
+  PermissionObjectSchema
+]);
 var AgentOverrideConfigSchema = z2.object({
   model: z2.union([
     z2.string(),
@@ -404,13 +460,15 @@ var AgentOverrideConfigSchema = z2.object({
   prompt: z2.string().min(1).optional(),
   orchestratorPrompt: z2.string().min(1).optional(),
   options: z2.record(z2.string(), z2.unknown()).optional(),
-  displayName: z2.string().min(1).optional()
+  displayName: z2.string().min(1).optional(),
+  permission: PermissionConfigSchema.optional()
 }).strict();
 var MultiplexerTypeSchema = z2.enum([
   "auto",
   "tmux",
   "zellij",
   "herdr",
+  "cmux",
   "none"
 ]);
 var MultiplexerLayoutSchema = z2.enum([
@@ -454,7 +512,9 @@ var FailoverConfigSchema = z2.object({
   enabled: z2.boolean().default(true),
   timeoutMs: z2.number().min(0).default(15000),
   retryDelayMs: z2.number().min(0).default(500),
-  retry_on_empty: z2.boolean().default(true).describe("When true (default), empty provider responses are treated as failures, " + "triggering fallback/retry. Set to false to treat them as successes.")
+  maxRetries: z2.number().int().min(0).default(3).describe("Number of consecutive 429/rate-limit responses tolerated on the " + "same model before aborting (or swapping to the next fallback " + "model when a chain is configured)."),
+  retry_on_empty: z2.boolean().default(true).describe("When true (default), empty provider responses are treated as failures, " + "triggering fallback/retry. Set to false to treat them as successes."),
+  runtimeOverride: z2.boolean().optional().describe("DEPRECATED: no longer used. Previously controlled whether out-of-chain " + "runtime model picks triggered fallback. Fallback is now always " + "disabled when a user explicitly selects a model via /model.")
 }).strict();
 var CompanionConfigSchema = z2.object({
   enabled: z2.boolean().optional(),
@@ -495,12 +555,14 @@ function rejectOrchestratorPromptOnOrchestrator(overrides, ctx, pathPrefix) {
 var PluginConfigSchema = z2.object({
   preset: z2.string().optional(),
   setDefaultAgent: z2.boolean().optional(),
-  compactSidebar: z2.boolean().optional().describe("Use the compact TUI sidebar layout when enabled."),
+  compactSidebar: z2.boolean().optional().describe("Use the compact TUI sidebar layout. Defaults to true; set false to use the expanded layout."),
+  stripOrchestratorModel: z2.boolean().optional().describe("When true, omit orchestrator.model and orchestrator.variant from the SDK config so OpenCode uses the session model selected with /model after subagent dispatch. An explicitly selected preset that sets orchestrator.model is preserved. Defaults to false."),
   autoUpdate: z2.boolean().optional().describe("Disable automatic installation of plugin updates when false. Defaults to true."),
   presets: z2.record(z2.string(), PresetSchema).optional(),
   agents: z2.record(z2.string(), AgentOverrideConfigSchema).optional(),
   disabled_agents: z2.array(z2.string()).optional().describe("Agent names to disable completely. " + "Disabled agents are not instantiated and cannot be delegated to. " + "Orchestrator and council internal agents (councillor) cannot be disabled. " + "By default, 'observer' is disabled. Remove it from this list and configure a vision-capable model to enable."),
-  disabled_mcps: z2.array(z2.string()).optional(),
+  image_routing: z2.enum(["auto", "direct"]).optional().describe("How image attachments are handled. " + "When omitted, preserves legacy conditional behavior: intercept " + 'attachments only when observer is enabled. "auto": requires ' + "observer to be enabled and saves attachments to disk before " + 'nudging delegation to @observer. "direct": always passes ' + "attachments to the orchestrator untouched."),
+  disabled_mcps: z2.array(z2.string()).optional().describe("MCP server names to disable completely. Disabled servers are not " + "started and cannot be used by agents."),
   disabled_tools: z2.array(z2.string()).optional().describe("Tool names to disable completely. Disabled tools are not registered with OpenCode and cannot be used by agents."),
   disabled_skills: z2.array(z2.string()).optional().describe("Skill names to disable completely. Disabled skills are not granted to agents, even when referenced by presets or agent overrides."),
   multiplexer: MultiplexerConfigSchema.optional(),
@@ -564,6 +626,12 @@ var CUSTOM_SKILLS = [
     allowedAgents: ["orchestrator"],
     sourcePath: "src/skills/deepwork"
   },
+  {
+    name: "verification-planning",
+    description: "Plan credible, proportionate evidence before non-trivial implementation",
+    allowedAgents: ["orchestrator"],
+    sourcePath: "src/skills/verification-planning"
+  },
   {
     name: "reflect",
     description: "Review repeated work and suggest reusable workflow improvements",
@@ -595,12 +663,12 @@ var SCHEMA_URL = "https://unpkg.com/oh-my-opencode-slim@latest/oh-my-opencode-sl
 var GENERATED_PRESETS = ["openai", "opencode-go"];
 var MODEL_MAPPINGS = {
   openai: {
-    orchestrator: { model: "openai/gpt-5.5", variant: "medium" },
-    oracle: { model: "openai/gpt-5.5", variant: "high" },
-    librarian: { model: "openai/gpt-5.4-mini", variant: "low" },
-    explorer: { model: "openai/gpt-5.4-mini", variant: "low" },
-    designer: { model: "openai/gpt-5.4-mini", variant: "medium" },
-    fixer: { model: "openai/gpt-5.5", variant: "low" }
+    orchestrator: { model: "openai/gpt-5.6-terra", variant: "medium" },
+    oracle: { model: "openai/gpt-5.6-sol", variant: "high" },
+    librarian: { model: "openai/gpt-5.6-luna", variant: "low" },
+    explorer: { model: "openai/gpt-5.6-luna", variant: "low" },
+    designer: { model: "openai/gpt-5.6-luna", variant: "medium" },
+    fixer: { model: "openai/gpt-5.6-luna", variant: "medium" }
   },
   kimi: {
     orchestrator: { model: "kimi-for-coding/k2p5" },
@@ -2726,7 +2794,9 @@ async function installCompanionArchive(finalBinaryPath, target, manifest, downlo
     if (tempDir) {
       try {
         rmSync3(tempDir, { recursive: true, force: true });
-      } catch {}
+      } catch (err) {
+        log("[updater] install cleanup failed", String(err));
+      }
     }
   }
 }
@@ -2736,7 +2806,9 @@ function readInstallMetadata(binaryPath) {
     if (parsed?.version && parsed.tag && parsed.target) {
       return parsed;
     }
-  } catch {}
+  } catch (err) {
+    log("[updater] metadata read failed", String(err));
+  }
   return null;
 }
 function writeInstallMetadata(binaryPath, metadata) {
@@ -2758,7 +2830,9 @@ async function withCompanionInstallLock(binaryPath, timeoutMs, staleMs, run) {
       } finally {
         try {
           rmSync3(lock, { recursive: true, force: true });
-        } catch {}
+        } catch (err) {
+          log("[updater] lock release failed", String(err));
+        }
       }
     } catch (err) {
       const code = err.code;