oh-my-opencode 4.18.2 → 4.19.0

package/README.ja.md

@@ -193,7 +193,7 @@ Read this and tell me why it's not just another boilerplate: https://raw.githubu
 <td align="center"><img src=".github/assets/hephaestus.png" height="300" /></td>
 </tr></table>
 
-**Sisyphus** (`claude-opus-4-7` / **`kimi-k2.6`** / **`glm-5`**) はあなたのメインオーケストレーターです。計画を立て、専門家に委任し、攻撃的な並列実行でタスクを完了まで推進します。途中で投げ出すことはありません。
+**Sisyphus** (`claude-opus-4-8` / `claude-opus-4-7` / **`kimi-k3`** / **`kimi-k2.6`** / **`glm-5`**) はあなたのメインオーケストレーターです。計画を立て、専門家に委任し、攻撃的な並列実行でタスクを完了まで推進します。途中で投げ出すことはありません。thinking token のコストを許容できる場合は **Kimi K3** を、そうでなければ **Claude Opus 4.8 / 4.7** または **Kimi K2.6** を推奨します。
 
 **Hephaestus** (OpenAI または Vercel では `gpt-5.6-sol`、その次に `gpt-5.5`) はあなたの自律的なディープワーカーです。レシピではなく、目標を与えてください。手取り足取り教えなくても、コードベースを探索し、パターンを調査し、エンドツーエンドで実行します。*正当なる職人 (The Legitimate Craftsman).*
 
@@ -203,7 +203,7 @@ Read this and tell me why it's not just another boilerplate: https://raw.githubu
 
 > Anthropic が [私たちのせいで OpenCode をブロックしました。](https://x.com/thdxr/status/2010149530486911014) だからこそ Hephaestus は「正当なる職人 (The Legitimate Craftsman)」と呼ばれているのです。皮肉を込めています。
 >
-> Opus で最もよく動きますが、Kimi K2.6 + GPT-5.5 の組み合わせだけでも、バニラの Claude Code を軽く凌駕します。設定は一切不要です。
+> Opus または Kimi K3 で最もよく動きますが、Kimi K2.6 + GPT-5.5 の組み合わせだけでも、バニラの Claude Code を軽く凌駕します。設定は一切不要です。
 
 ### Team Mode (v4.0)
 

package/README.ko.md

@@ -194,7 +194,7 @@ Read this and tell me why it's not just another boilerplate: https://raw.githubu
 <td align="center"><img src=".github/assets/hephaestus.png" height="300" /></td>
 </tr></table>
 
-**Sisyphus** (`claude-opus-4-7` / **`kimi-k2.6`** / **`glm-5`**)는 메인 오케스트레이터입니다. 계획을 세우고, 전문가에게 위임하고, 공격적인 병렬 실행으로 작업을 끝까지 밀어붙입니다. 중간에 멈추지 않습니다.
+**Sisyphus** (`claude-opus-4-8` / `claude-opus-4-7` / **`kimi-k3`** / **`kimi-k2.6`** / **`glm-5`**)는 메인 오케스트레이터입니다. 계획을 세우고, 전문가에게 위임하고, 공격적인 병렬 실행으로 작업을 끝까지 밀어붙입니다. 중간에 멈추지 않습니다. thinking token 비용을 감수할 수 있다면 **Kimi K3**를, 그렇지 않으면 **Claude Opus 4.8 / 4.7** 또는 **Kimi K2.6**을 권장합니다.
 
 **Hephaestus** (OpenAI 또는 Vercel에서 `gpt-5.6-sol`, 그다음 `gpt-5.5`)는 자율적으로 깊게 파는 작업자입니다. 레시피가 아니라 목표를 주세요. 코드베이스를 탐색하고, 패턴을 조사하고, 손을 잡아주지 않아도 엔드투엔드로 실행합니다. *The Legitimate Craftsman.*
 
@@ -204,7 +204,7 @@ Read this and tell me why it's not just another boilerplate: https://raw.githubu
 
 > Anthropic은 [우리 때문에 OpenCode를 차단했습니다.](https://x.com/thdxr/status/2010149530486911014) 그래서 Hephaestus에게 "The Legitimate Craftsman"이라는 별명이 붙었습니다. 의도된 아이러니입니다.
 >
-> Opus에서 가장 잘 돌지만, Kimi K2.6 + GPT-5.5 조합만으로도 이미 바닐라 Claude Code를 이깁니다. 별도 설정 없이요.
+> Opus나 Kimi K3에서 가장 잘 돌지만, Kimi K2.6 + GPT-5.5 조합만으로도 이미 바닐라 Claude Code를 이깁니다. 별도 설정 없이요.
 
 ### Team Mode (v4.0)
 

package/README.md

@@ -221,7 +221,7 @@ Even with only the following subscriptions, `ultrawork` works well (this project
 |   🔎   | **AST-Grep**                                             | Ultimate | Pattern-aware code search and rewriting across 25 languages.                                                                                                                                                     |
 |   🧠   | **Background Agents**                                    | Ultimate | Fire 5+ specialists in parallel. Context stays lean. Results when ready.                                                                                                                                         |
 |   📚   | **Built-in MCPs** (web/docs/code search)                 | Ultimate | Exa (web search), Context7 (official docs), Grep.app (GitHub search). Always on.                                                                                                                                 |
-|   🔁   | **Ralph Loop / `/ulw-loop`**                             | Ultimate | Self-referential loop. Doesn't stop until 100% done.                                                                                                                                                             |
+|   🔁   | **Goal / `/goal`**                                       | Ultimate | Persistent per-session objective. Re-injects a continuation prompt on every idle until a completion audit says it's done.                                                                                        |
 |   ✅   | **Todo Enforcer**                                        | Ultimate | Agent goes idle? System yanks it back. Your task gets done, period.                                                                                                                                              |
 |   💬   | **Comment Checker**                                      | Both     | No AI slop in comments. Code reads like a senior wrote it.                                                                                                                                                       |
 |   📐   | **Rules Injection** (`AGENTS.md` / `.omo/rules/**`)      | Both     | Project rules and AGENTS.md auto-loaded into the agent's context at every prompt.                                                                                                                                |
@@ -241,7 +241,7 @@ Even with only the following subscriptions, `ultrawork` works well (this project
 <td align="center"><img src=".github/assets/hephaestus.png" height="300" /></td>
 </tr></table>
 
-**Sisyphus** (`claude-opus-4-7` / **`kimi-k2.6`** / **`glm-5`** ) is your main orchestrator. He plans, delegates to specialists, and drives tasks to completion with aggressive parallel execution. He does not stop halfway.
+**Sisyphus** (`claude-opus-4-8` / `claude-opus-4-7` / **`kimi-k3`** / **`kimi-k2.6`** / **`glm-5`** ) is your main orchestrator. He plans, delegates to specialists, and drives tasks to completion with aggressive parallel execution. He does not stop halfway. Use **Kimi K3** when you can accept its thinking-token cost; otherwise **Claude Opus 4.8 / 4.7** or **Kimi K2.6** are the recommended defaults.
 
 **Hephaestus** (`gpt-5.6-sol` on OpenAI or Vercel, then `gpt-5.5`) is your autonomous deep worker. Give him a goal, not a recipe. He explores the codebase, researches patterns, and executes end-to-end without hand-holding. *The Legitimate Craftsman.*
 
@@ -251,7 +251,7 @@ Every agent is tuned to its model's specific strengths. No manual model juggling
 
 > Anthropic [blocked OpenCode because of us.](https://x.com/thdxr/status/2010149530486911014) That's why Hephaestus is called "The Legitimate Craftsman." The irony is intentional.
 >
-> We run best on Opus, but Kimi K2.6 + GPT-5.5 already beats vanilla Claude Code. Zero config needed.
+> We run best on Opus or Kimi K3, but Kimi K2.6 + GPT-5.5 already beats vanilla Claude Code. Zero config needed.
 
 ### Team Mode (v4.0)
 
@@ -433,7 +433,7 @@ See full [Features Documentation](docs/reference/features.md).
 - **Claude Code Compatibility**: Full hook system, commands, skills, agents, MCPs
 - **Built-in MCPs**: websearch (Exa), context7 (docs), grep_app (GitHub search) — injected at runtime by the plugin; not visible in `opencode mcp list` (see [MCP docs](docs/reference/features.md#native-vs-plugin-injected-mcps))
 - **Session Tools**: List, read, search, and analyze session history
-- **Productivity Features**: Ralph Loop, Todo Enforcer, Comment Checker, Think Mode, and more
+- **Productivity Features**: Goal, Todo Enforcer, Comment Checker, Think Mode, and more
 - **Doctor Command**: Built-in diagnostics (`bunx oh-my-opencode doctor`) verify plugin registration, config, models, and environment
 - **Model Fallbacks**: `fallback_models` can mix plain model strings with per-fallback object settings in the same array
 - **File Prompts**: Load prompts from files with `file://` support in agent configurations

package/README.ru.md

@@ -192,7 +192,7 @@ Read this and tell me why it's not just another boilerplate: https://raw.githubu
 <td align="center"><img src=".github/assets/hephaestus.png" height="300" /></td>
 </tr></table>
 
-**Sisyphus** (`claude-opus-4-7` / **`kimi-k2.6`** / **`glm-5`**) — главный оркестратор. Он планирует, делегирует задачи специалистам и доводит их до завершения с агрессивным параллельным выполнением. Он не останавливается на полпути.
+**Sisyphus** (`claude-opus-4-8` / `claude-opus-4-7` / **`kimi-k3`** / **`kimi-k2.6`** / **`glm-5`**) — главный оркестратор. Он планирует, делегирует задачи специалистам и доводит их до завершения с агрессивным параллельным выполнением. Он не останавливается на полпути. Используйте **Kimi K3**, если готовы оплатить повышенный расход thinking-токенов; иначе рекомендуем **Claude Opus 4.8 / 4.7** или **Kimi K2.6**.
 
 **Hephaestus** (`gpt-5.6-sol` через OpenAI или Vercel, затем `gpt-5.5`) — автономный глубокий исполнитель. Дайте ему цель, а не рецепт. Он исследует кодовую базу, изучает паттерны и выполняет задачи сквозным образом без лишних подсказок. *Законный Мастер.*
 
@@ -202,7 +202,7 @@ Read this and tell me why it's not just another boilerplate: https://raw.githubu
 
 > Anthropic [заблокировал OpenCode из-за нас.](https://x.com/thdxr/status/2010149530486911014) Именно поэтому Hephaestus зовётся «Законным Мастером». Ирония намеренная.
 >
-> Мы работаем лучше всего на Opus, но Kimi K2.6 + GPT-5.5 уже превосходят ванильный Claude Code. Никакой настройки не требуется.
+> Мы работаем лучше всего на Opus или Kimi K3, но Kimi K2.6 + GPT-5.5 уже превосходят ванильный Claude Code. Никакой настройки не требуется.
 
 ### Team Mode (v4.0)
 

package/README.zh-cn.md

@@ -199,7 +199,7 @@ Read this and tell me why it's not just another boilerplate: https://raw.githubu
 <td align="center"><img src=".github/assets/hephaestus.png" height="300" /></td>
 </tr></table>
 
-**Sisyphus** (`claude-opus-4-7` / **`kimi-k2.6`** / **`glm-5`**) 是你的主指挥官。他负责制定计划、分配任务给专家团队，并以极其激进的并行策略推动任务直至完成。他从不半途而废。
+**Sisyphus** (`claude-opus-4-8` / `claude-opus-4-7` / **`kimi-k3`** / **`kimi-k2.6`** / **`glm-5`**) 是你的主指挥官。他负责制定计划、分配任务给专家团队，并以极其激进的并行策略推动任务直至完成。他从不半途而废。如果你能接受 thinking token 的成本，请使用 **Kimi K3**；否则推荐使用 **Claude Opus 4.8 / 4.7** 或 **Kimi K2.6**。
 
 **Hephaestus** (在 OpenAI 或 Vercel 上优先 `gpt-5.6-sol`，其次 `gpt-5.5`) 是你的自主深度工作者。你只需要给他目标，不要给他具体做法。他会自动探索代码库模式，从头到尾独立执行任务，绝不会中途要你当保姆。*名副其实的正牌工匠。*
 
@@ -209,7 +209,7 @@ Read this and tell me why it's not just another boilerplate: https://raw.githubu
 
 > Anthropic [因为我们屏蔽了 OpenCode](https://x.com/thdxr/status/2010149530486911014)。这就是为什么我们将 Hephaestus 命名为"正牌工匠 (The Legitimate Craftsman)"。这是一个故意的讽刺。
 >
-> 我们在 Opus 上运行得最好，但仅仅使用 Kimi K2.6 + GPT-5.5 就足以碾压原版的 Claude Code。完全不需要配置。
+> 我们在 Opus 或 Kimi K3 上运行得最好，但仅仅使用 Kimi K2.6 + GPT-5.5 就足以碾压原版的 Claude Code。完全不需要配置。
 
 ### Team Mode (v4.0)
 

package/dist/agents/atlas/agent.d.ts

@@ -7,16 +7,17 @@
  * 1. Claude Opus 4.7       → opus-4-7.md   (literal-following + explicit fan-out push)
  * 2. GPT family            → gpt.md        (calibrated for GPT-5.5)
  * 3. Gemini family         → gemini.md
- * 4. Kimi K2.7             → kimi-k2-7.md  (restrained, outcome-first; checked before generic kimi)
- * 5. Kimi K2.x family      → kimi.md       (Claude-family base + K2.6 thinking-mode calibration)
- * 6. GLM family            → glm.md        (GLM 5.2 calibration)
- * 7. Default (Claude 4.6 family: opus-4-6, sonnet-4-6, haiku-4-5, etc.) → default.md
+ * 4. Kimi K3              → kimi-k3.md    (K3-native; checked first among Kimi)
+ * 5. Kimi K2.7             → kimi-k2-7.md  (restrained, outcome-first; checked before generic kimi)
+ * 6. Kimi K2.x family      → kimi.md       (Claude-family base + K2.6 thinking-mode calibration)
+ * 7. GLM family            → glm.md        (GLM 5.2 calibration)
+ * 8. Default (Claude 4.6 family: opus-4-6, sonnet-4-6, haiku-4-5, etc.) → default.md
  */
 import type { AgentConfig } from "@opencode-ai/sdk";
 import type { AgentPromptMetadata } from "../types";
 import type { AvailableAgent, AvailableSkill } from "../dynamic-agent-prompt-builder";
 import type { CategoryConfig } from "../../config/schema";
-export type AtlasPromptSource = "default" | "gpt" | "gemini" | "kimi" | "kimi-k2-7" | "opus-4-7" | "glm";
+export type AtlasPromptSource = "default" | "gpt" | "gemini" | "kimi" | "kimi-k3" | "kimi-k2-7" | "opus-4-7" | "glm";
 export declare function getAtlasPromptSource(model?: string): AtlasPromptSource;
 export interface OrchestratorContext {
     model?: string;

package/dist/agents/sisyphus/index.d.ts

@@ -9,6 +9,9 @@
  * - gemini.ts: Corrective overlays for Gemini's aggressive tendencies
  * - gpt-5-4.ts: Native GPT-5.4 prompt with block-structured guidance
  * - gpt-5-5.ts: Native GPT-5.5 prompt with Codex-style sections
+ * - kimi-k2-6.ts: Kimi K2.6 native prompt with thinking-mode stop conditions
+ * - kimi-k2-7.ts: Kimi K2.7 native prompt, restrained and outcome-first
+ * - kimi-k3.ts: Kimi K3 native prompt with explicit anti-overthinking calibration
  */
 export { buildDefaultSisyphusPrompt, buildTaskManagementSection } from "./default";
 export { buildClaudeOpus47SisyphusPrompt } from "./claude-opus-4-7";
@@ -19,3 +22,5 @@ export { buildGpt54SisyphusPrompt } from "./gpt-5-4";
 export { buildGpt55SisyphusPrompt } from "./gpt-5-5";
 export { buildGlm52SisyphusPrompt } from "./glm-5-2";
 export { buildKimiK26SisyphusPrompt } from "./kimi-k2-6";
+export { buildKimiK27SisyphusPrompt } from "./kimi-k2-7";
+export { buildKimiK3SisyphusPrompt } from "./kimi-k3";

package/dist/agents/sisyphus/kimi-k3.d.ts

@@ -0,0 +1,19 @@
+/**
+ * Kimi K3-native Sisyphus prompt.
+ *
+ * K3 inherits K2.7's restrained, outcome-first steerability, but its thinking
+ * policy is tuned even harder for long-horizon reasoning. The resulting value
+ * is real on ambiguous architecture, multi-step debugging, and failure
+ * diagnosis. The cost is a tendency to keep thinking once the decisive
+ * condition is already met — re-deriving given facts, enumerating unused
+ * alternatives, and second-guessing settled choices.
+ *
+ * This prompt does not suppress reasoning. It gives the reasoning a stop
+ * condition: terminal conditions, commitment framing, and an explicit
+ * "go work" rule. Everything else is the same shared builder surface as the
+ * other Sisyphus variants.
+ */
+import type { AvailableAgent, AvailableTool, AvailableSkill, AvailableCategory } from "../dynamic-agent-prompt-builder";
+import { categorizeTools } from "../dynamic-agent-prompt-builder";
+export declare function buildKimiK3SisyphusPrompt(model: string, availableAgents: AvailableAgent[], availableTools?: AvailableTool[], availableSkills?: AvailableSkill[], availableCategories?: AvailableCategory[], useTaskSystem?: boolean): string;
+export { categorizeTools };

package/dist/agents/sisyphus-agent-factory.d.ts

@@ -7,7 +7,7 @@ import type { AvailableAgent, AvailableCategory, AvailableSkill } from "./dynami
  * runtime model resolves to a different family than the configured one, the baked
  * body is the wrong family and must be rebuilt (issue #5297/#5316).
  */
-export type SisyphusPromptFamily = "kimi-k2-7" | "kimi-k2-6" | "gpt-5-5" | "gpt-5-4" | "claude-fable-5" | "claude-opus-4-8" | "claude-opus-4-7" | "glm-5-2" | "fallback";
+export type SisyphusPromptFamily = "kimi-k3" | "kimi-k2-7" | "kimi-k2-6" | "gpt-5-5" | "gpt-5-4" | "claude-fable-5" | "claude-opus-4-8" | "claude-opus-4-7" | "glm-5-2" | "fallback";
 export declare function resolveSisyphusPromptFamily(model: string): SisyphusPromptFamily;
 export declare function createSisyphusAgent(model: string, availableAgents?: AvailableAgent[], availableToolNames?: string[], availableSkills?: AvailableSkill[], availableCategories?: AvailableCategory[], useTaskSystem?: boolean): AgentConfig;
 export declare namespace createSisyphusAgent {

package/dist/agents/sisyphus-junior/agent.d.ts

@@ -5,9 +5,13 @@
  * Category-spawned executor with domain-specific configurations.
  *
  * Routing:
- * 1. GPT models (openai/*, github-copilot/gpt-*) -> gpt.ts (GPT-5.4 optimized)
- * 2. Gemini models (google/*, google-vertex/*) -> gemini.ts (Gemini-optimized)
- * 3. Default (Claude, etc.) -> default.ts (Claude-optimized)
+ * 1. Kimi K3 -> kimi-k3.ts (K3-native executor; reasoning depth with built-in stop conditions)
+ * 2. Kimi K2.7 -> kimi-k2-7.ts (restrained, outcome-first)
+ * 3. Kimi K2.x -> kimi-k2-6.ts
+ * 4. GPT models (openai/*, github-copilot/gpt-*) -> gpt-5-5.ts / gpt-5-4.ts / gpt.ts
+ * 5. Gemini models (google/*, google-vertex/*) -> gemini.ts (Gemini-optimized)
+ * 6. GLM models -> glm-5-2.ts
+ * 7. Default (Claude, etc.) -> default.ts (Claude-optimized)
  */
 import type { AgentConfig } from "@opencode-ai/sdk";
 import type { AgentOverrideConfig } from "../../config/schema";
@@ -15,7 +19,7 @@ export declare const SISYPHUS_JUNIOR_DEFAULTS: {
     readonly model: "anthropic/claude-sonnet-4-6";
     readonly temperature: 0.1;
 };
-export type SisyphusJuniorPromptSource = "default" | "kimi-k2" | "kimi-k2-7" | "gpt" | "gpt-5-5" | "gpt-5-4" | "gemini" | "glm-5-2";
+export type SisyphusJuniorPromptSource = "default" | "kimi-k2" | "kimi-k2-7" | "kimi-k3" | "gpt" | "gpt-5-5" | "gpt-5-4" | "gemini" | "glm-5-2";
 export declare function getSisyphusJuniorPromptSource(model?: string): SisyphusJuniorPromptSource;
 /**
  * Builds the appropriate Sisyphus-Junior prompt based on model.

package/dist/agents/sisyphus-junior/index.d.ts

@@ -1,5 +1,7 @@
 export { buildDefaultSisyphusJuniorPrompt } from "./default";
 export { buildKimiK26SisyphusJuniorPrompt } from "./kimi-k2-6";
+export { buildKimiK27SisyphusJuniorPrompt } from "./kimi-k2-7";
+export { buildKimiK3SisyphusJuniorPrompt } from "./kimi-k3";
 export { buildGptSisyphusJuniorPrompt } from "./gpt";
 export { buildGpt54SisyphusJuniorPrompt } from "./gpt-5-4";
 export { buildGpt55SisyphusJuniorPrompt } from "./gpt-5-5";

package/dist/agents/sisyphus-junior/kimi-k3.d.ts

@@ -0,0 +1,12 @@
+/**
+ * Kimi K3-native Sisyphus-Junior prompt.
+ *
+ * Authored for K3 as a complete prompt, not a K2.7 tune with a calibration
+ * appendix. K3 keeps K2.7's restrained, outcome-first steerability and pushes
+ * the thinking policy harder toward long-horizon reasoning. On a focused
+ * executor that depth pays off in multi-step debugging and failure diagnosis;
+ * left unshaped it keeps reasoning after the decisive condition is met. The
+ * stop conditions are therefore woven into the sections where they act — the
+ * role, the task read, the tool loop — with verification rigor first-class.
+ */
+export declare function buildKimiK3SisyphusJuniorPrompt(useTaskSystem: boolean, promptAppend?: string): string;

package/dist/agents/types.d.ts

@@ -1,6 +1,6 @@
 import type { AgentConfig } from "@opencode-ai/sdk";
-import { isClaudeFable5Model, isClaudeFableOrMythosModel, isClaudeOpus46Model, isClaudeOpus47Model, isClaudeOpus47OrLaterModel, isClaudeOpus48Model, isGeminiModel, isGlmModel, isGptModel, isKimiK2Model, isKimiK27Model, isMiniMaxModel } from "@oh-my-opencode/model-core";
-export { isClaudeFable5Model, isClaudeFableOrMythosModel, isClaudeOpus46Model, isClaudeOpus47Model, isClaudeOpus47OrLaterModel, isClaudeOpus48Model, isGeminiModel, isGlmModel, isGptModel, isKimiK2Model, isKimiK27Model, isMiniMaxModel, };
+import { isClaudeFable5Model, isClaudeFableOrMythosModel, isClaudeOpus46Model, isClaudeOpus47Model, isClaudeOpus47OrLaterModel, isClaudeOpus48Model, isGeminiModel, isGlmModel, isGptModel, isKimiK2Model, isKimiK27Model, isKimiK3Model, isMiniMaxModel } from "@oh-my-opencode/model-core";
+export { isClaudeFable5Model, isClaudeFableOrMythosModel, isClaudeOpus46Model, isClaudeOpus47Model, isClaudeOpus47OrLaterModel, isClaudeOpus48Model, isGeminiModel, isGlmModel, isGptModel, isKimiK2Model, isKimiK27Model, isKimiK3Model, isMiniMaxModel, };
 /**
  * Anthropic Opus 4.7+, Fable, and Mythos models reject thinking.type "enabled";
  * they require adaptive thinking plus an effort, which OpenCode core derives from