release: v0.1.16

* fix(publish): make git operations idempotent

- Check for staged changes before commit
- Check if tag exists before creating
- Check if release exists before creating

* fix(session-recovery): handle empty content from interrupted reasoning

- Add THINKING_TYPES set including 'reasoning' type (OpenCode's thinking)
- Add hasNonEmptyOutput() to detect messages with only thinking/meta parts
- Add findEmptyContentMessage() to scan all messages for empty content
- Handle step-start/step-finish meta parts in empty content detection
- Patch interrupted messages with '(interrupted)' text before falling back to revert

AGENTS.md

@@ -84,9 +84,31 @@ bun run build
 bun run rebuild
 ```
 
+## DEPLOYMENT
+
+**배포는 GitHub Actions workflow_dispatch로만 진행**
+
+1. package.json 버전은 수정하지 않음 (워크플로우에서 자동 bump)
+2. 변경사항 커밋 & 푸시
+3. GitHub Actions에서 `publish` 워크플로우 수동 실행
+   - `bump`: major | minor | patch 선택
+   - `version`: (선택) 특정 버전 지정 가능
+
+```bash
+# 워크플로우 실행 (CLI)
+gh workflow run publish -f bump=patch
+
+# 워크플로우 상태 확인
+gh run list --workflow=publish
+```
+
+**주의사항**:
+- `bun publish` 직접 실행 금지 (OIDC provenance 문제)
+- 로컬에서 버전 bump 하지 말 것
+
 ## NOTES
 
 - **No tests**: Test framework not configured
-- **No CI/CD**: GitHub workflows not present
+- **CI/CD**: GitHub Actions publish workflow 사용
 - **Version requirement**: OpenCode >= 1.0.132 (earlier versions have config bugs)
 - **Multi-language docs**: README.md, README.en.md, README.ko.md

README.ko.md

@@ -96,76 +96,9 @@ OpenCode 가 낭만이 사라진것같은 오늘날의 시대에, 당신에게
 
 ### Agent 설정
 
-각 에이전트의 모델, 프롬프트, 권한 등을 세밀하게 조정할 수 있어.
+`agents` 옵션으로 에이전트의 모델, 프롬프트, 권한 등을 세밀하게 조정할 수 있습니다. 스키마를 통해 자동완성을 지원합니다.
 
-**설정 옵션:**
-
-| 옵션 | 설명 |
-|------|------|
-| `model` | 사용할 모델 ID (예: `anthropic/claude-sonnet-4`) |
-| `temperature` | 창의성 조절 (0.0 ~ 2.0) |
-| `top_p` | 단어 선택 다양성 (0.0 ~ 1.0) |
-| `prompt` | 시스템 프롬프트 오버라이드 |
-| `tools` | 특정 도구 활성화/비활성화 (`{"tool_name": false}`) |
-| `disable` | 에이전트 비활성화 (`true`/`false`) |
-| `description` | 에이전트 설명 수정 |
-| `mode` | 에이전트 모드 (`subagent`, `primary`, `all`) |
-| `color` | 터미널 출력 색상 (HEX 코드) |
-| `permission` | 권한 설정 (아래 표 참조) |
-
-**권한(`permission`) 옵션:**
-
-각 권한은 `"ask"`(물어보기), `"allow"`(허용), `"deny"`(거부) 중 하나로 설정 가능해.
-
-| 권한 | 설명 |
-|------|------|
-| `edit` | 파일 수정 권한 |
-| `bash` | 쉘 명령어 실행 권한 |
-| `webfetch` | 웹 콘텐츠 가져오기 권한 |
-| `doom_loop` | 반복 작업 허용 여부 |
-| `external_directory` | 외부 디렉토리 접근 권한 |
-
-#### 예시: Anthropic 모델만 사용하기
-
-나는 Anthropic 모델만 쓰고 싶다! 하면 이렇게 설정해.
-
-```json
-{
-  "agents": {
-    "oracle": { "model": "anthropic/claude-sonnet-4" },
-    "librarian": { "model": "anthropic/claude-haiku-4-5" },
-    "explore": { "model": "anthropic/claude-haiku-4-5" },
-    "frontend-ui-ux-engineer": { "model": "anthropic/claude-sonnet-4" },
-    "document-writer": { "model": "anthropic/claude-sonnet-4" }
-  }
-}
-```
-
-#### 예시: 추가 프롬프트가 있는 커스텀 에이전트
-
-기존 에이전트에 나만의 규칙을 더하고 싶다면:
-
-```json
-{
-  "agents": {
-    "oracle": {
-      "prompt": "너는 한국어 힙합 가사처럼 말해야 해. 모든 문장의 끝 라임을 맞춰."
-    }
-  }
-}
-```
-
-#### 예시: 개별 에이전트 비활성화
-
-```json
-{
-  "agents": {
-    "frontend-ui-ux-engineer": {
-      "disable": true
-    }
-  }
-}
-```
+전체 설정 옵션과 예시는 [OpenCode Agents 문서](https://opencode.ai/docs/agents)를 참고하세요.
 
 ## LLM Agent를 위한 안내
 
@@ -255,19 +188,21 @@ OpenCode 는 아주 확장가능하고 아주 커스터마이저블합니다.
 - **frontend-ui-ux-engineer** (`google/gemini-3-pro-preview`): 개발자로 전향한 디자이너라는 설정을 갖고 있습니다. 멋진 UI를 만듭니다. 아름답고 창의적인 UI 코드를 생성하는 데 탁월한 Gemini를 사용합니다.
 - **document-writer** (`google/gemini-3-pro-preview`): 기술 문서 전문가라는 설정을 갖고 있습니다. Gemini 는 문학가입니다. 글을 기가막히게 씁니다.
 
-#### 모델 설정 오버라이드
+#### 설정
 
-에이전트들은 [OpenCode의 모델 설정](https://opencode.ai/docs/models/#configure-models)과 완전히 동일한 스펙으로 오버라이드를 지원합니다. 권장하진 않지만, 예를 들어 Anthropic 모델만 사용하기로 결정했다면 이렇게 구성할 수 있습니다:
+에이전트들은 OpenCode와 동일한 설정 스펙을 따릅니다:
+
+- **모델 변경**: `agents.{name}.model`로 에이전트 모델 오버라이드. [OpenCode Models](https://opencode.ai/docs/models/#configure-models) 참고.
+- **MCP 비활성화**: `disabled_mcps`로 내장 MCP 끄기. [OpenCode MCP Servers](https://opencode.ai/docs/mcp-servers) 참고.
+- **에이전트 비활성화**: `disabled_agents` 또는 `agents.{name}.disable` 사용. [OpenCode Agents](https://opencode.ai/docs/agents) 참고.
+
+권장하진 않지만(이 플러그인은 멀티 모델 오케스트레이션용), Anthropic만 사용하는 경우 예시:
 
 ```json
 {
   "agents": {
-    "explore": {
-      "model": "anthropic/claude-haiku-4-5"
-    },
-    "frontend-ui-ux-engineer": {
-      "model": "anthropic/claude-opus-4"
-    }
+    "explore": { "model": "anthropic/claude-haiku-4-5" },
+    "frontend-ui-ux-engineer": { "model": "anthropic/claude-opus-4" }
   }
 }
 ```

README.md

@@ -146,89 +146,9 @@ Available agents: `oracle`, `librarian`, `explore`, `frontend-ui-ux-engineer`, `
 
 ### Agent Configuration
 
-You can override the configuration of any built-in agent using the `agents` option. This allows you to change models, adjust creativity, modify permissions, or disable agents individually.
+You can override any built-in agent's model, prompt, permissions, and more using the `agents` option. Configuration uses autocomplete via schema.
 
-#### Configuration Options
-
-| Option | Type | Description |
-|--------|------|-------------|
-| `model` | string | Override the default model (e.g., "anthropic/claude-sonnet-4") |
-| `temperature` | number (0-2) | Controls randomness (0 = deterministic, 2 = creative) |
-| `top_p` | number (0-1) | Nucleus sampling parameter |
-| `prompt` | string | Additional system prompt to append |
-| `tools` | object | Enable/disable specific tools (e.g., `{"websearch_exa": false}`) |
-| `disable` | boolean | Completely disable the agent |
-| `description` | string | Override agent description |
-| `mode` | "subagent" | "primary" | "all" | When agent is available |
-| `color` | string | Hex color code for terminal output (e.g., "#FF0000") |
-| `permission` | object | Permission settings for sensitive operations |
-
-#### Permission Options
-
-| Option | Values | Description |
-|--------|--------|-------------|
-| `edit` | "ask" | "allow" | "deny" | File modification permissions |
-| `bash` | "ask" | "allow" | "deny" | object | Shell command execution permissions |
-| `webfetch` | "ask" | "allow" | "deny" | Web access permissions |
-| `doom_loop` | "ask" | "allow" | "deny" | Infinite loop prevention |
-| `external_directory` | "ask" | "allow" | "deny" | Access outside project root |
-
-#### Examples
-
-**Using Only Anthropic Models**
-
-This configuration forces all agents to use Anthropic models, suitable for users with only Anthropic API access.
-
-```json
-{
-  "agents": {
-    "oracle": {
-      "model": "anthropic/claude-sonnet-4"
-    },
-    "librarian": {
-      "model": "anthropic/claude-haiku-4-5"
-    },
-    "explore": {
-      "model": "anthropic/claude-haiku-4-5"
-    },
-    "frontend-ui-ux-engineer": {
-      "model": "anthropic/claude-sonnet-4"
-    },
-    "document-writer": {
-      "model": "anthropic/claude-sonnet-4"
-    }
-  }
-}
-```
-
-**Custom Agent with Additional Prompt**
-
-Inject custom instructions into an agent's system prompt.
-
-```json
-{
-  "agents": {
-    "frontend-ui-ux-engineer": {
-      "prompt": "ALWAYS use Tailwind CSS. NEVER use inline styles. Prefer dark mode defaults.",
-      "temperature": 0.8
-    }
-  }
-}
-```
-
-**Disable Agents Individually**
-
-You can also disable agents using the `disable` property within the agent config.
-
-```json
-{
-  "agents": {
-    "explore": {
-      "disable": true
-    }
-  }
-}
-```
+For full configuration options and examples, see the [OpenCode Agents documentation](https://opencode.ai/docs/agents).
 
 ## Why OpenCode & Why Oh My OpenCode
 
@@ -265,19 +185,21 @@ I believe in the right tool for the job. For your wallet's sake, use CLIProxyAPI
 - **frontend-ui-ux-engineer** (`google/gemini-3-pro-preview`): A designer turned developer. Creates stunning UIs. Uses Gemini because its creativity and UI code generation are superior.
 - **document-writer** (`google/gemini-3-pro-preview`): A technical writing expert. Gemini is a wordsmith; it writes prose that flows naturally.
 
-#### Model Configuration Override
+#### Configuration
 
-Agents follow the exact same model configuration spec as [OpenCode's model configuration](https://opencode.ai/docs/models/#configure-models). While not generally recommended, if you decide to use only Anthropic models, you could configure like this:
+Agents follow the same configuration spec as OpenCode:
+
+- **Change models**: Override any agent's model via `agents.{name}.model`. See [OpenCode Models](https://opencode.ai/docs/models/#configure-models).
+- **Disable MCPs**: Use `disabled_mcps` to turn off built-in MCPs. See [OpenCode MCP Servers](https://opencode.ai/docs/mcp-servers).
+- **Disable agents**: Use `disabled_agents` or `agents.{name}.disable`. See [OpenCode Agents](https://opencode.ai/docs/agents).
+
+While not generally recommended (this plugin is designed for multi-model orchestration), here's an example for Anthropic-only users:
 
 ```json
 {
   "agents": {
-    "explore": {
-      "model": "anthropic/claude-haiku-4-5"
-    },
-    "frontend-ui-ux-engineer": {
-      "model": "anthropic/claude-opus-4"
-    }
+    "explore": { "model": "anthropic/claude-haiku-4-5" },
+    "frontend-ui-ux-engineer": { "model": "anthropic/claude-opus-4" }
   }
 }
 ```

package.json

@@ -1,6 +1,6 @@
 {
   "name": "oh-my-opencode",
-  "version": "0.1.14",
+  "version": "0.1.16",
   "description": "OpenCode plugin - custom agents (oracle, librarian) and enhanced features",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",

script/publish.ts

@@ -78,13 +78,35 @@ async function gitTagAndRelease(newVersion: string, changelog: string): Promise<
   await $`git config user.email "github-actions[bot]@users.noreply.github.com"`
   await $`git config user.name "github-actions[bot]"`
   await $`git add package.json`
-  await $`git commit -m "release: v${newVersion}"`
-  await $`git tag v${newVersion}`
+
+  // Commit only if there are staged changes (idempotent)
+  const hasStagedChanges = await $`git diff --cached --quiet`.nothrow()
+  if (hasStagedChanges.exitCode !== 0) {
+    await $`git commit -m "release: v${newVersion}"`
+  } else {
+    console.log("No changes to commit (version already updated)")
+  }
+
+  // Tag only if it doesn't exist (idempotent)
+  const tagExists = await $`git rev-parse v${newVersion}`.nothrow()
+  if (tagExists.exitCode !== 0) {
+    await $`git tag v${newVersion}`
+  } else {
+    console.log(`Tag v${newVersion} already exists`)
+  }
+
+  // Push (idempotent - git push is already idempotent)
   await $`git push origin HEAD --tags`
 
+  // Create release only if it doesn't exist (idempotent)
   console.log("\nCreating GitHub release...")
   const releaseNotes = changelog || "No notable changes"
-  await $`gh release create v${newVersion} --title "v${newVersion}" --notes ${releaseNotes}`
+  const releaseExists = await $`gh release view v${newVersion}`.nothrow()
+  if (releaseExists.exitCode !== 0) {
+    await $`gh release create v${newVersion} --title "v${newVersion}" --notes ${releaseNotes}`
+  } else {
+    console.log(`Release v${newVersion} already exists`)
+  }
 }
 
 async function checkVersionExists(version: string): Promise<boolean> {

src/hooks/session-recovery.ts

@@ -211,29 +211,93 @@ async function recoverThinkingDisabledViolation(
   return false
 }
 
+const THINKING_TYPES = new Set(["thinking", "redacted_thinking", "reasoning"])
+
+function hasNonEmptyOutput(msg: MessageData): boolean {
+  const parts = msg.parts
+  if (!parts || parts.length === 0) return false
+
+  return parts.some((p) => {
+    if (THINKING_TYPES.has(p.type)) return false
+    if (p.type === "step-start" || p.type === "step-finish") return false
+    if (p.type === "text" && p.text && p.text.trim()) return true
+    if (p.type === "tool_use" && p.id) return true
+    if (p.type === "tool_result") return true
+    return false
+  })
+}
+
+function findEmptyContentMessage(msgs: MessageData[]): MessageData | null {
+  for (let i = 0; i < msgs.length; i++) {
+    const msg = msgs[i]
+    const isLastMessage = i === msgs.length - 1
+    const isAssistant = msg.info?.role === "assistant"
+
+    if (isLastMessage && isAssistant) continue
+
+    if (!hasNonEmptyOutput(msg)) {
+      return msg
+    }
+  }
+  return null
+}
+
 async function recoverEmptyContentMessage(
   client: Client,
   sessionID: string,
   failedAssistantMsg: MessageData,
   directory: string
 ): Promise<boolean> {
-  const messageID = failedAssistantMsg.info?.id
-  const parentMsgID = failedAssistantMsg.info?.parentID
-
-  if (!messageID) {
-    return false
-  }
-
-  // Revert to parent message (delete the empty message)
-  const revertTargetID = parentMsgID || messageID
-
   try {
+    const messagesResp = await client.session.messages({
+      path: { id: sessionID },
+      query: { directory },
+    })
+    const msgs = (messagesResp as { data?: MessageData[] }).data
+
+    if (!msgs || msgs.length === 0) return false
+
+    const emptyMsg = findEmptyContentMessage(msgs) || failedAssistantMsg
+    const messageID = emptyMsg.info?.id
+    if (!messageID) return false
+
+    const existingParts = emptyMsg.parts || []
+    const hasOnlyThinkingOrMeta = existingParts.length > 0 && existingParts.every(
+      (p) => THINKING_TYPES.has(p.type) || p.type === "step-start" || p.type === "step-finish"
+    )
+
+    if (hasOnlyThinkingOrMeta) {
+      const strippedParts: MessagePart[] = [{ type: "text", text: "(interrupted)" }]
+
+      try {
+        // @ts-expect-error - Experimental API
+        await client.message?.update?.({
+          path: { id: messageID },
+          body: { parts: strippedParts },
+        })
+        return true
+      } catch {
+        // message.update not available
+      }
+
+      try {
+        // @ts-expect-error - Experimental API
+        await client.session.patch?.({
+          path: { id: sessionID },
+          body: { messageID, parts: strippedParts },
+        })
+        return true
+      } catch {
+        // session.patch not available
+      }
+    }
+
+    const revertTargetID = emptyMsg.info?.parentID || messageID
     await client.session.revert({
       path: { id: sessionID },
       body: { messageID: revertTargetID },
       query: { directory },
     })
-
     return true
   } catch {
     return false

src/tools/ast-grep/tools.ts

@@ -10,13 +10,55 @@ function showOutputToUser(context: unknown, output: string): void {
   ctx.metadata?.({ metadata: { output } })
 }
 
+/**
+ * JS/TS languages that require complete function declaration patterns
+ */
+const JS_TS_LANGUAGES = ["javascript", "typescript", "tsx"] as const
+
+/**
+ * Validates AST pattern for common incomplete patterns that will fail silently.
+ * Only validates JS/TS languages where function declarations require body.
+ *
+ * @throws Error with helpful message if pattern is incomplete
+ */
+function validatePatternForCli(pattern: string, lang: CliLanguage): void {
+  if (!JS_TS_LANGUAGES.includes(lang as (typeof JS_TS_LANGUAGES)[number])) {
+    return
+  }
+
+  const src = pattern.trim()
+
+  // Detect incomplete function declarations:
+  // - "function $NAME" (no params/body)
+  // - "export function $NAME" (no params/body)
+  // - "export async function $NAME" (no params/body)
+  // - "export default function $NAME" (no params/body)
+  // Pattern: ends with $METAVAR (uppercase, underscore, digits) without ( or {
+  const incompleteFunctionDecl =
+    /^(export\s+)?(default\s+)?(async\s+)?function\s+\$[A-Z_][A-Z0-9_]*\s*$/i.test(src)
+
+  if (incompleteFunctionDecl) {
+    throw new Error(
+      `Incomplete AST pattern for ${lang}: "${pattern}"\n\n` +
+        `ast-grep requires complete AST nodes. Function declarations must include parameters and body.\n\n` +
+        `Examples of correct patterns:\n` +
+        `  - "export async function $NAME($$$) { $$$ }" (matches export async functions)\n` +
+        `  - "function $NAME($$$) { $$$ }" (matches all function declarations)\n` +
+        `  - "async function $NAME($$$) { $$$ }" (matches async functions)\n\n` +
+        `Your pattern "${pattern}" is missing the parameter list and body.`
+    )
+  }
+}
+
 export const ast_grep_search = tool({
   description:
     "Search code patterns across filesystem using AST-aware matching. Supports 25 languages. " +
     "Use meta-variables: $VAR (single node), $$$ (multiple nodes). " +
+    "IMPORTANT: Patterns must be complete AST nodes (valid code). " +
+    "For functions, include params and body: 'export async function $NAME($$$) { $$$ }' not 'export async function $NAME'. " +
     "Examples: 'console.log($MSG)', 'def $FUNC($$$):', 'async function $NAME($$$)'",
   args: {
-    pattern: tool.schema.string().describe("AST pattern with meta-variables ($VAR, $$$)"),
+    pattern: tool.schema.string().describe("AST pattern with meta-variables ($VAR, $$$). Must be complete AST node."),
     lang: tool.schema.enum(CLI_LANGUAGES).describe("Target language"),
     paths: tool.schema.array(tool.schema.string()).optional().describe("Paths to search (default: ['.'])"),
     globs: tool.schema.array(tool.schema.string()).optional().describe("Include/exclude globs (prefix ! to exclude)"),
@@ -24,6 +66,8 @@ export const ast_grep_search = tool({
   },
   execute: async (args, context) => {
     try {
+      validatePatternForCli(args.pattern, args.lang as CliLanguage)
+
       const matches = await runSg({
         pattern: args.pattern,
         lang: args.lang as CliLanguage,