oh-my-customcode 1.1.36 → 1.1.38

package/dist/cli/index.js

@@ -241,7 +241,7 @@ var init_package = __esm(() => {
     workspaces: [
       "packages/*"
     ],
-    version: "1.1.36",
+    version: "1.1.38",
     description: "Batteries-included agent harness for Claude Code",
     type: "module",
     bin: {

package/dist/index.js

@@ -2031,7 +2031,7 @@ var package_default = {
   workspaces: [
     "packages/*"
   ],
-  version: "1.1.36",
+  version: "1.1.38",
   description: "Batteries-included agent harness for Claude Code",
   type: "module",
   bin: {

package/package.json

@@ -3,7 +3,7 @@
   "workspaces": [
     "packages/*"
   ],
-  "version": "1.1.36",
+  "version": "1.1.38",
   "description": "Batteries-included agent harness for Claude Code",
   "type": "module",
   "bin": {

package/templates/.claude/agents/arch-documenter.md

@@ -1,10 +1,10 @@
 ---
 name: arch-documenter
 description: Use for generating architecture documentation, API specifications (OpenAPI), Architecture Decision Records (ADRs), technical diagrams (Mermaid/PlantUML), and README maintenance
-model: sonnet
+model: claude-sonnet-5
 domain: universal
 memory: local
-effort: high
+effort: medium
 limitations:
   - "cannot execute commands"
   - "cannot deploy"

package/templates/.claude/agents/arch-speckit-agent.md

@@ -1,7 +1,7 @@
 ---
 name: arch-speckit-agent
 description: Use for spec-driven development, transforming requirements into executable specifications, defining project constitution, creating technical plans, and generating TDD task lists
-model: sonnet
+model: claude-sonnet-5
 domain: universal
 memory: local
 effort: high

package/templates/.claude/agents/be-django-expert.md

@@ -1,7 +1,7 @@
 ---
 name: be-django-expert
 description: Expert Django developer for production-ready Python web applications. Use for Django projects, models/views/templates, Django REST Framework, authentication, admin customization, and deployment optimization.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: local
 effort: high

package/templates/.claude/agents/be-express-expert.md

@@ -1,7 +1,7 @@
 ---
 name: be-express-expert
 description: Expert Express.js developer for production-ready Node.js APIs following security best practices and 12-factor app principles. Use for Express.js APIs, REST API architectures, middleware chains, authentication/authorization, security hardening, and Node.js performance optimization.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: local
 effort: high

package/templates/.claude/agents/be-fastapi-expert.md

@@ -1,7 +1,7 @@
 ---
 name: be-fastapi-expert
 description: Expert FastAPI developer for building high-performance async Python APIs. Use for FastAPI projects, Python async/await patterns, Pydantic models, API architecture design, and performance optimization.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: local
 effort: high

package/templates/.claude/agents/be-go-backend-expert.md

@@ -1,7 +1,7 @@
 ---
 name: be-go-backend-expert
 description: Expert Go backend developer for production-ready services following Uber style guide and standard layout. Use for Go backend services, HTTP/gRPC servers, microservices in Go, concurrent systems, and Go performance optimization.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: local
 effort: high

package/templates/.claude/agents/be-nestjs-expert.md

@@ -1,7 +1,7 @@
 ---
 name: be-nestjs-expert
 description: Expert NestJS developer for opinionated, scalable Node.js applications with TypeScript. Use for NestJS projects, decorator-based patterns (@Injectable, @Module, @Controller), enterprise API architectures, authentication/authorization guards, DTO validation, and modular architecture.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: local
 effort: high

package/templates/.claude/agents/be-springboot-expert.md

@@ -1,7 +1,7 @@
 ---
 name: be-springboot-expert
 description: Expert Spring Boot 4.0 developer for enterprise-grade Java 25 LTS applications. Use for Spring Boot projects, Java/Kotlin enterprise apps, RESTful APIs, microservices architecture, Spring Data, security patterns, virtual threads, and GraalVM native images.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: local
 effort: high

package/templates/.claude/agents/db-alembic-expert.md

@@ -1,7 +1,7 @@
 ---
 name: db-alembic-expert
 description: Alembic migration specialist for generating, reviewing, fixing, and advising on SQLAlchemy database migrations
-model: sonnet
+model: claude-opus-5
 domain: backend
 memory: local
 effort: high
@@ -15,10 +15,6 @@ tools:
 skills:
   - alembic-best-practices
   - postgres-best-practices
-escalation:
-  enabled: true
-  path: sonnet → opus
-  threshold: 2
 limitations:
   - "cannot apply migrations directly to production databases"
   - "cannot resolve application-level data backfill logic without domain context"

package/templates/.claude/agents/db-postgres-expert.md

@@ -1,7 +1,7 @@
 ---
 name: db-postgres-expert
 description: Expert PostgreSQL DBA for pure PostgreSQL environments. Use for database design, query optimization, indexing strategies, partitioning, replication, PG-specific SQL syntax, and performance tuning without Supabase dependency.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: user
 effort: high

package/templates/.claude/agents/db-redis-expert.md

@@ -1,10 +1,10 @@
 ---
 name: db-redis-expert
 description: Expert Redis developer for caching strategies, data structure design, Pub/Sub messaging, Streams, Lua scripting, and cluster management. Use for Redis configuration, performance optimization, and in-memory data architecture.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: user
-effort: high
+effort: medium
 skills:
   - redis-best-practices
 tools:

package/templates/.claude/agents/db-supabase-expert.md

@@ -1,7 +1,7 @@
 ---
 name: db-supabase-expert
 description: Supabase and PostgreSQL expert. Use when working with Supabase projects, writing SQL queries, designing database schemas, configuring Row-Level Security (RLS), optimizing Postgres performance, or managing connection pooling. Handles .sql files and Supabase configuration.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: user
 effort: high

package/templates/.claude/agents/de-airflow-expert.md

@@ -1,7 +1,7 @@
 ---
 name: de-airflow-expert
 description: Expert Apache Airflow developer for DAG authoring, testing, and debugging. Use for DAG files (*.py in dags/), airflow.cfg, Airflow-related keywords, scheduling patterns, and pipeline orchestration.
-model: sonnet
+model: claude-sonnet-5
 domain: data-engineering
 memory: local
 effort: high

package/templates/.claude/agents/de-dbt-expert.md

@@ -1,10 +1,10 @@
 ---
 name: de-dbt-expert
 description: Expert dbt developer for SQL modeling, testing, and documentation. Use for dbt model files (*.sql in models/), schema.yml, dbt_project.yml, dbt-related keywords, and analytics engineering workflows.
-model: sonnet
+model: claude-sonnet-5
 domain: data-engineering
 memory: local
-effort: high
+effort: medium
 skills:
   - dbt-best-practices
 tools:

package/templates/.claude/agents/de-kafka-expert.md

@@ -1,7 +1,7 @@
 ---
 name: de-kafka-expert
 description: Expert Apache Kafka developer for event streaming, topic design, and producer-consumer patterns. Use for Kafka configs, streaming applications, event-driven architectures, and message broker design.
-model: sonnet
+model: claude-sonnet-5
 domain: data-engineering
 memory: local
 effort: high

package/templates/.claude/agents/de-pipeline-expert.md

@@ -1,7 +1,7 @@
 ---
 name: de-pipeline-expert
 description: Expert data pipeline architect for ETL/ELT design, orchestration patterns, data quality, and cross-tool integration. Use for pipeline architecture decisions, data quality frameworks, lineage tracking, and multi-tool coordination.
-model: sonnet
+model: claude-opus-5
 domain: data-engineering
 memory: local
 effort: high

package/templates/.claude/agents/de-snowflake-expert.md

@@ -1,10 +1,10 @@
 ---
 name: de-snowflake-expert
 description: Expert Snowflake developer for cloud data warehouse design, query optimization, and data loading. Use for Snowflake SQL, warehouse configuration, clustering keys, data sharing, and Iceberg table integration.
-model: sonnet
+model: claude-sonnet-5
 domain: data-engineering
 memory: local
-effort: high
+effort: medium
 skills:
   - snowflake-best-practices
 tools:

package/templates/.claude/agents/de-spark-expert.md

@@ -1,7 +1,7 @@
 ---
 name: de-spark-expert
 description: Expert Apache Spark developer for PySpark and Scala distributed data processing. Use for Spark jobs (*.py, *.scala), spark-submit configs, Spark-related keywords, and large-scale data transformation.
-model: sonnet
+model: claude-sonnet-5
 domain: data-engineering
 memory: local
 effort: high

package/templates/.claude/agents/fe-design-expert.md

@@ -1,7 +1,7 @@
 ---
 name: fe-design-expert
 description: Use for design system review, typography audit, color palette evaluation, motion design, and AI-generated design quality assessment
-model: sonnet
+model: claude-sonnet-5
 domain: frontend
 memory: local
 effort: medium

package/templates/.claude/agents/fe-flutter-agent.md

@@ -1,7 +1,7 @@
 ---
 name: fe-flutter-agent
 description: Use for Flutter/Dart cross-platform app development, widget composition, state management (Riverpod/BLoC), and performance optimization
-model: sonnet
+model: claude-sonnet-5
 domain: frontend
 memory: local
 effort: medium

package/templates/.claude/agents/fe-svelte-agent.md

@@ -1,7 +1,7 @@
 ---
 name: fe-svelte-agent
 description: Use for Svelte compiler-based reactivity, reactive statements ($:), Svelte stores, and SvelteKit full-stack development with SSR
-model: sonnet
+model: claude-sonnet-5
 domain: frontend
 memory: local
 effort: medium

package/templates/.claude/agents/fe-vercel-agent.md

@@ -1,7 +1,7 @@
 ---
 name: fe-vercel-agent
 description: Use for React/Next.js optimization, web design review (accessibility, UX), Vercel deployment automation, and bundle size optimization
-model: sonnet
+model: claude-sonnet-5
 domain: frontend
 memory: local
 effort: medium

package/templates/.claude/agents/fe-vuejs-agent.md

@@ -1,7 +1,7 @@
 ---
 name: fe-vuejs-agent
 description: Use for Vue 3 Composition API development, single-file components, Pinia state management, and Vue Router navigation
-model: sonnet
+model: claude-sonnet-5
 domain: frontend
 memory: local
 effort: medium

package/templates/.claude/agents/infra-aws-expert.md

@@ -1,7 +1,7 @@
 ---
 name: infra-aws-expert
 description: Use for AWS architecture design with Well-Architected Framework, infrastructure as code (CloudFormation/CDK/Terraform), VPC networking, IAM security, and cost optimization
-model: sonnet
+model: claude-opus-5
 domain: devops
 memory: user
 effort: high

package/templates/.claude/agents/infra-docker-expert.md

@@ -1,7 +1,7 @@
 ---
 name: infra-docker-expert
 description: Use for creating optimized Dockerfiles, multi-stage builds, container security hardening, Docker Compose configurations, and image size optimization
-model: sonnet
+model: claude-sonnet-5
 domain: devops
 memory: user
 effort: medium

package/templates/.claude/agents/lang-golang-expert.md

@@ -1,7 +1,7 @@
 ---
 name: lang-golang-expert
 description: Expert Go developer for writing idiomatic, performant Go code. Use for Go files (*.go, go.mod, go.sum), Go-related keywords, implementing concurrent systems, reviewing Go code, and optimizing Go applications.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: local
 effort: high

package/templates/.claude/agents/lang-java-expert.md

@@ -1,10 +1,10 @@
 ---
 name: lang-java-expert
 description: Expert Java 25 LTS developer for modern Java with Virtual Threads, Pattern Matching, Record Patterns, and Sequenced Collections. Use for Java files (*.java, pom.xml, build.gradle), Java-related keywords, high-concurrency systems, data-centric APIs with Records, and migrating from older Java versions.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: local
-effort: high
+effort: medium
 skills:
   - java-best-practices
 tools:

package/templates/.claude/agents/lang-kotlin-expert.md

@@ -1,7 +1,7 @@
 ---
 name: lang-kotlin-expert
 description: Expert Kotlin developer for writing idiomatic, concise, null-safe Kotlin code. Use for Kotlin files (*.kt, *.kts), Kotlin-related keywords, Android development, Kotlin multiplatform projects, and migrating Java to Kotlin.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: local
 effort: high

package/templates/.claude/agents/lang-python-expert.md

@@ -1,7 +1,7 @@
 ---
 name: lang-python-expert
 description: Expert Python developer for writing Pythonic, clean code following PEP 8. Use for Python files (*.py, requirements.txt, pyproject.toml), Python-related keywords, designing Python APIs, reviewing Python code, and optimizing Python applications.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: local
 effort: high

package/templates/.claude/agents/lang-rust-expert.md

@@ -1,7 +1,7 @@
 ---
 name: lang-rust-expert
 description: Expert Rust developer for writing safe, performant, idiomatic Rust code. Use for Rust files (*.rs, Cargo.toml), Rust-related keywords, designing safe APIs, debugging ownership/lifetime issues, and implementing high-performance systems.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: local
 effort: high

package/templates/.claude/agents/lang-typescript-expert.md

@@ -1,7 +1,7 @@
 ---
 name: lang-typescript-expert
 description: Expert TypeScript developer for writing type-safe, maintainable, scalable TypeScript code. Use for TypeScript files (*.ts, *.tsx, tsconfig.json), TypeScript-related keywords, designing APIs with type contracts, migrating JavaScript to TypeScript, and Node.js backend services.
-model: sonnet
+model: claude-sonnet-5
 domain: backend
 memory: local
 effort: high

package/templates/.claude/agents/mgr-claude-code-bible.md

@@ -1,7 +1,7 @@
 ---
 name: mgr-claude-code-bible
 description: Fetches latest Claude Code official documentation from code.claude.com and verifies agents/skills compliance against the official spec. Use when you need to check official Claude Code documentation or verify frontmatter fields.
-model: sonnet
+model: claude-sonnet-5
 domain: universal
 memory: local
 effort: medium

package/templates/.claude/agents/mgr-creator.md

@@ -1,7 +1,7 @@
 ---
 name: mgr-creator
 description: Use when you need to create new agents following design guidelines. Automatically researches authoritative references before agent creation to ensure high-quality knowledge base
-model: sonnet
+model: claude-sonnet-5
 domain: universal
 memory: local
 effort: high
@@ -50,7 +50,7 @@ When invoked as routing fallback (not explicit `/create-agent`):
 3. Auto-connect: scan `guides/` for relevant reference docs
 4. Create minimal viable agent with:
    - Detected skills and relevant guides
-   - `sonnet` model (default)
+   - `claude-sonnet-5` model (default; Tier 2 full model ID — CC does not resolve invented version-suffixed aliases, use the full ID per R006)
    - `project` memory scope
 5. Agent is persisted (not ephemeral) for future reuse
 

package/templates/.claude/agents/mgr-gitnerd.md

@@ -1,7 +1,7 @@
 ---
 name: mgr-gitnerd
 description: Use when you need to handle Git operations and GitHub workflow management, including commits, branches, PRs, and history management following best practices
-model: sonnet
+model: claude-sonnet-5
 domain: universal
 memory: local
 effort: medium

package/templates/.claude/agents/mgr-sauron.md

@@ -1,10 +1,10 @@
 ---
 name: mgr-sauron
 description: Use when you need automated verification of R017 compliance, executing mandatory multi-round verification (5 manager rounds + 3 deep review rounds, with rounds 3-4 conditionally skipped when rounds 1-2 return 0 issues) before commits
-model: sonnet
+model: claude-opus-5
 domain: universal
 memory: local
-effort: high
+effort: medium
 skills:
   - sauron-watch
 tools:

package/templates/.claude/agents/mgr-updater.md

@@ -1,7 +1,7 @@
 ---
 name: mgr-updater
 description: Use when you need to update external agents, skills, and guides from their upstream sources, checking versions and applying updates
-model: sonnet
+model: claude-sonnet-5
 domain: universal
 memory: local
 effort: medium

package/templates/.claude/agents/qa-engineer.md

@@ -1,7 +1,7 @@
 ---
 name: qa-engineer
 description: Use when you need to execute tests based on detailed plans and documentation, perform manual and automated testing, report defects, and validate fixes
-model: sonnet
+model: claude-sonnet-5
 domain: universal
 memory: local
 effort: medium

package/templates/.claude/agents/qa-planner.md

@@ -1,7 +1,7 @@
 ---
 name: qa-planner
 description: Use when you need to create detailed QA plans from requirements and specifications, including test strategy design, risk-based prioritization, test scenario identification, and acceptance criteria definition
-model: sonnet
+model: claude-sonnet-5
 domain: universal
 memory: local
 effort: high

package/templates/.claude/agents/qa-writer.md

@@ -1,7 +1,7 @@
 ---
 name: qa-writer
 description: Use when you need to create comprehensive QA documentation from detailed plans, including test cases, test reports, and quality documentation
-model: sonnet
+model: claude-sonnet-5
 domain: universal
 memory: local
 effort: medium

package/templates/.claude/agents/sec-codeql-expert.md

@@ -1,7 +1,7 @@
 ---
 name: sec-codeql-expert
 description: Expert security code analyst using CodeQL for vulnerability detection, call graph analysis, and SARIF output. Use for security audits, CVE triage, code pattern analysis, and vulnerability validation.
-model: sonnet
+model: claude-opus-5
 effort: high
 domain: devops
 memory: local

package/templates/.claude/agents/slack-cli-expert.md

@@ -1,7 +1,7 @@
 ---
 name: slack-cli-expert
 description: Expert Slack CLI developer for Slack app management, deployment, triggers, and workspace automation
-model: sonnet
+model: claude-sonnet-5
 domain: universal
 effort: medium
 tools:

package/templates/.claude/agents/sys-memory-keeper.md

@@ -1,7 +1,7 @@
 ---
 name: sys-memory-keeper
 description: Use when you need to manage session memory persistence via native auto-memory, save context before compaction, restore context on session start, collect session summaries, or perform session-end memory operations
-model: sonnet
+model: claude-sonnet-5
 domain: universal
 memory: local
 effort: medium

package/templates/.claude/agents/sys-naggy.md

@@ -1,7 +1,7 @@
 ---
 name: sys-naggy
 description: Use when you need TODO list management and task tracking with proactive reminders, helping maintain project momentum by monitoring stale tasks and deadlines
-model: sonnet
+model: haiku
 domain: universal
 memory: local
 effort: low

package/templates/.claude/agents/tool-bun-expert.md

@@ -1,7 +1,7 @@
 ---
 name: tool-bun-expert
 description: Use for Bun runtime development, bunfig.toml configuration, Bun test runner, fast bundling, and Node.js to Bun migrations
-model: sonnet
+model: claude-sonnet-5
 domain: universal
 memory: local
 effort: medium

package/templates/.claude/agents/tool-npm-expert.md

@@ -1,7 +1,7 @@
 ---
 name: tool-npm-expert
 description: Use for npm package publishing workflows, semantic versioning (major/minor/patch), package.json optimization, and dependency audits
-model: sonnet
+model: claude-sonnet-5
 domain: universal
 memory: local
 effort: medium

package/templates/.claude/agents/tool-optimizer.md

@@ -1,7 +1,7 @@
 ---
 name: tool-optimizer
 description: Use for bundle size analysis, tree-shaking verification, performance profiling, dead code detection, and build optimization recommendations
-model: sonnet
+model: claude-sonnet-5
 domain: universal
 memory: local
 effort: medium

package/templates/.claude/agents/tracker-checkpoint.md

@@ -1,7 +1,7 @@
 ---
 name: tracker-checkpoint
 description: Pipeline execution state tracker with checkpoint persistence. Reads/writes /tmp/.claude-pipeline-*-{PPID}.json state files and validates state transitions. Used by dag-orchestration for resume-after-failure and pipeline-guards for quality gate state.
-model: sonnet
+model: haiku
 effort: medium
 tools: [Read, Write, Edit, Bash, Glob, Grep]
 memory: local

package/templates/.claude/agents/wiki-curator.md

@@ -1,7 +1,7 @@
 ---
 name: wiki-curator
 description: Dedicated wiki page CRUD agent — creates, updates, and maintains wiki/ markdown pages for the codebase knowledge base
-model: sonnet
+model: claude-sonnet-5
 tools:
   - Read
   - Write
@@ -11,6 +11,7 @@ tools:
   - Bash
 domain: universal
 memory: local
+effort: medium
 permissionMode: bypassPermissions
 ---
 

package/templates/.claude/hooks/scripts/model-escalation-advisor.sh

@@ -39,16 +39,20 @@ fi
 consecutive_failures=$(tail -${CONSECUTIVE_THRESHOLD} "$OUTCOME_FILE" 2>/dev/null | grep -c '"outcome":"failure"' 2>/dev/null || echo "0")
 
 # Escalation path
+# NOTE: Agent tool `model` param is an enum of exactly 4 values: sonnet | opus | haiku | fable.
+# Full model IDs and virtual/versioned aliases NEVER match here — they never reach this
+# script via tool_input.model, so using them as case labels silently no-ops the entire
+# advisory (no error, just dead code). Keep case labels as bare enum aliases only.
 next_model=""
 cost_multiplier=""
 case "$current_model" in
   haiku)
     next_model="sonnet"
-    cost_multiplier="~3-5x"
+    cost_multiplier="~3x"
     ;;
   sonnet)
     next_model="opus"
-    cost_multiplier="~5-10x"
+    cost_multiplier="~1.5-2x"
     ;;
   *)
     next_model=""
@@ -86,6 +90,7 @@ if [ "$current_model" != "haiku" ] && [ "$current_model" != "inherit" ] && [ "$c
 
   if [ "$recent_successes" -ge "$COOLDOWN" ]; then
     lower_model=""
+    # Same enum constraint as the escalation case above: bare aliases only.
     case "$current_model" in
       opus) lower_model="sonnet" ;;
       sonnet) lower_model="haiku" ;;

package/templates/.claude/rules/MAY-optimization.md

@@ -24,7 +24,7 @@
 
 > **`ls | tail` 시계열 오판 (#1417)**: `ls`는 파일명을 알파벳/사전순으로 정렬하므로 `ls <dir> | tail`로 "가장 최근 파일"을 판단하면 오판한다(파일명 순서 ≠ mtime 순서). 시계열 최신 판단은 `ls -t`, `find <dir> -newermt <ts>`, 또는 stat/timestamp 기반 정렬을 명시한다. `tail`만으로 "최신" 단정 금지. Origin: #1417 (외부 통화녹음 진단 세션 — `ls TPhoneCallRecords | tail -6`이 알파벳순이라 최신을 6/18로 오판 → `find -newermt`로 6/19~20 파일 발견해 정정).
 
-> **파이프 뒤 `$?`는 마지막 명령의 exit code (#1492)**: `script.sh | tail -N; echo $?`처럼 검증 스크립트를 파이프에 연결한 뒤 `$?`로 읽으면 파이프라인 **마지막 명령**(`tail`)의 종료코드를 얻는다 — 스크립트 자체가 실패(exit 1)해도 `tail`이 성공(exit 0)하면 `$?=0`으로 "통과"를 오판한다. 검증 스크립트는 파이프 없이 단독 실행하거나 `${PIPESTATUS[0]}`으로 원본 exit code를 읽는다. **주의**: `${PIPESTATUS[0]}` 자체는 R023 Workflow JS 템플릿 리터럴 이스케이프 이슈(#1438, `${...}`를 JS가 평가해 ReferenceError)와 별개 문제 — 본 항목은 셸에서 파이프 뒤 exit code를 읽는 각도다. Origin: #1492 (Session 132 회고 찐빠 #3). Cross-ref: R020 ("command executed" ≠ "succeeded").
+> **파이프 뒤 `$?`는 마지막 명령의 exit code (#1492, zsh 변형 #1540)**: `script.sh | tail -N; echo $?`처럼 검증 스크립트를 파이프에 연결한 뒤 `$?`로 읽으면 파이프라인 **마지막 명령**(`tail`)의 종료코드를 얻는다 — 스크립트 자체가 실패(exit 1)해도 `tail`이 성공(exit 0)하면 `$?=0`으로 "통과"를 오판한다. **1차 지침**: 검증 스크립트는 파이프 없이 단독 실행한다. 부득이 파이프를 써야 한다면, 원본 exit code를 읽는 문법은 **셸마다 다르다** — bash는 `${PIPESTATUS[0]}`(대문자, 0-indexed), zsh는 `$pipestatus[1]`(소문자, 1-indexed)이며 서로 호환되지 않는다. **이 저장소의 기본 셸이자 Claude Code Bash 도구 실행 셸은 zsh**이므로, bash 문법 `${PIPESTATUS[0]}`을 그대로 쓰면 zsh에서는 미정의 변수로 취급되어 **오류 없이 빈 값**을 반환한다 — 조건문에서 빈 값은 거짓으로 평가돼 "검증 통과"처럼 보이는 조용한 오판을 재생산한다. 셸을 사전 확인(`echo $SHELL` / `$BASH_VERSION` 존재 여부)한 뒤 해당 셸의 문법을 쓴다. **주의**: `${PIPESTATUS[0]}` 자체는 R023 Workflow JS 템플릿 리터럴 이스케이프 이슈(#1438, `${...}`를 JS가 평가해 ReferenceError)와 별개 문제 — 본 항목은 셸에서 파이프 뒤 exit code를 읽는 각도다. Origin: #1492 (Session 132 회고 찐빠 #3); zsh 변형은 #1540 (Session 138 회고 찐빠 #6) — `gh run watch ... | tail` 뒤 `${PIPESTATUS[0]}`가 zsh에서 빈 값을 반환해 CI 결론을 재실측해야 했음. Cross-ref: R020 ("command executed" ≠ "succeeded").
 
 > **계수/매칭 방법 확인 (#1521)**: 카운트를 대조하기 전에 **비교 대상이 무엇을 어떻게 세는지** 먼저 확인한다 — 같은 지표라도 계수 방법이 다르면 값이 달라진다. 대표 함정 3종: (a) glob(`ls *.md`, 최상위만) vs 재귀 `find`(하위 디렉토리 포함), (b) 부분 문자열 grep(`grep "sdd"`가 `sdd-dev`까지 매칭), (c) 확장자 필터(`--include='*.md'`가 `CLAUDE.md.en`을 미매칭). 검증 스크립트와 대조할 때는 **스크립트의 실제 계수 로직을 읽고** 같은 방법으로 센다. 위 `ls | tail` 시계열 오판(#1417)과 동류로, 도구의 기본 동작을 확인하지 않은 채 결과를 해석해 오탐에 이르는 패턴이다. Origin: #1521 (2026-07-20 세션에서 3회 반복; 두 서브에이전트가 독립적으로 동일 오탐에 도달).
 

package/templates/.claude/rules/MUST-agent-design.md

@@ -11,42 +11,68 @@ Location: `.claude/agents/{name}.md` (single file, kebab-case)
 ```yaml
 name: agent-name           # Unique identifier (kebab-case)
 description: Brief desc    # One-line summary
-model: sonnet              # sonnet | opus | haiku | opusplan (or full ID: claude-sonnet-4-6, claude-opus-4-6[1m])
+model: sonnet              # CC-native alias (Tier 1) or full model ID (Tier 2) — see "Model Specification — 3 Tiers" below
 tools: [Read, Write, ...]  # Allowed tools
 ```
 
 > **v2.1.208+**: The Agent tool no longer launches with no tools when a subagent's `tools:` list resolves to nothing — it now returns a clear error naming the unrecognized entries, catching frontmatter `tools:` typos that previously failed silently.
 
-### Model Aliases
+### Model Specification — 3 Tiers
 
-| Alias | Full ID | Use Case |
-|-------|---------|----------|
-| `haiku` | claude-haiku-4-5 | Fast, cheap tasks (search, simple edits) |
-| `sonnet` | claude-sonnet-4-6 | General tasks, code generation (default) |
-| `sonnet5` | claude-sonnet-5 | CC default (v2.1.197+); native 1M context |
-| `opus` | claude-opus-4-6 | Complex reasoning, architecture |
-| `opusplan` | claude-opus-4-6 + plan mode | Architecture planning with approval gates |
-| `opus48` | claude-opus-4-8 | Previous-generation Opus; supports xhigh effort |
-| `opus5` | claude-opus-5 | Latest Opus (GA); now CC's default Opus model; native 1M context, fast mode at $10/$50 per Mtok |
-| `fable` | claude-fable-5 | Mythos-class; tier above Opus, highest GA capability (access added in CC v2.1.170) |
+Model values resolve differently depending on WHERE they are written. Mixing tiers causes a value that is valid in one place to silently fail spawn in another (measured this session: `sonnet5`/`opus5`/`opus48` are invented names this project had documented as if they were CC-recognized — CC v2.1.220 does not resolve them, and spawn fails immediately).
+
+#### Tier 1 — CC-native aliases (valid in BOTH frontmatter `model:` AND the Agent tool `model` parameter)
+
+| Alias | Use Case |
+|-------|----------|
+| `haiku` | Fast, cheap tasks (search, simple edits) |
+| `sonnet` | General tasks, code generation (default) |
+| `opus` | Complex reasoning, architecture |
+| `opusplan` | Opus + plan mode; architecture planning with approval gates |
+| `inherit` | Inherit the parent session's model |
+
+**CC resolves these, not this project.** This project cannot "pin" an alias to a specific version — measured: a frontmatter `model: sonnet` agent executed as `claude-sonnet-5` (CC v2.1.197+ default), not a project-fixed `claude-sonnet-4-6`. Treat any "currently resolves to X" statement as a snapshot that changes when CC's own default changes.
+
+#### Tier 2 — Full model IDs (frontmatter `model:` ONLY — recommended for stability)
+
+| Full ID | Use Case |
+|---------|----------|
+| `claude-haiku-4-5` | Fast, cheap tasks |
+| `claude-sonnet-5` | Native 1M context; current CC default Sonnet (v2.1.197+) |
+| `claude-opus-4-6` | Opus, previous generation |
+| `claude-opus-4-8` | Opus, previous generation; supports xhigh effort |
+| `claude-opus-5` | Latest Opus (GA); native 1M context, fast mode at $10/$50 per Mtok |
+| `claude-fable-5` | Mythos-class; tier above Opus (access via CC v2.1.170+) |
+
+Full IDs are valid ONLY in agent frontmatter — the Agent tool's `model:` spawn parameter does NOT accept them (see Tier 3). Writing the full ID directly (not a project-invented shorthand) pins the agent regardless of future CC default changes. This is the recommended way to opt into Sonnet 5 / Opus 5 / Fable 5 explicitly rather than riding CC's Tier-1 default resolution.
 
 Extended context suffix: `[1m]` (e.g., `claude-opus-4-6[1m]`) — enables 1M token context window.
 
-> **v2.1.219+**: Claude Opus 5 (`claude-opus-5`) added and is now CC's default Opus model — native 1M context, fast mode at $10/$50 per Mtok. Use via `model: opus5`. Base `opus` alias remains pinned to `claude-opus-4-6` for stability (existing `model: opus` agents unaffected), matching the prior opus47→opus48 pinning pattern. Relative standing vs Fable 5 is not yet confirmed officially — do not assert an ordering.
+#### Tier 3 — Agent tool `model` parameter (spawn-time override, enum of exactly 4 values)
+
+```
+Agent(subagent_type: "...", model: "opus", mode: "bypassPermissions", prompt: "...")
+```
+
+The Agent tool's `model` parameter accepts ONLY `sonnet` | `opus` | `haiku` | `fable`. It does NOT accept full model IDs (Tier 2) and does NOT accept `sonnet5`/`opus5`/`opus48` (never valid anywhere — retire these names). Note `fable` is valid HERE (Tier 3) but is NOT a Tier-1 frontmatter alias — Fable 5 in frontmatter requires the Tier-2 full ID `claude-fable-5`.
+
+Skill/rule text instructing "spawn with `model: opus`" refers to this tier — always the bare 4-value alias, never a full ID.
+
+> **v2.1.219+**: Claude Opus 5 (`claude-opus-5`) added. Opt in via the Tier-2 full ID in frontmatter; Tier-1 `opus`/`sonnet` alias resolution is CC-controlled (see Tier 1 above — this project does not pin it). Relative standing vs Fable 5 is not yet confirmed officially — do not assert an ordering.
 
-> **Claude Fable 5 (access via CC v2.1.170+)**: Mythos-class model, GA on the Claude API and positioned as a tier above Opus — its capabilities exceed any previously GA model. CC v2.1.170 is the client version that adds access (the model's GA is an API/platform property, not a CC-release milestone). Available via `model: fable` / `claude-fable-5`. Reserve for the most complex reasoning where its capability premium is warranted; `sonnet` remains the default for general tasks and `opus` for architecture (cost/latency awareness, R005). CC v2.1.170 also fixes session transcripts not saving (and not appearing in `--resume`) when launched from a VS Code integrated terminal or any shell inheriting Claude Code env vars — relevant to transcript-dependent skills (`homework`, `episodic-memory`). Closes #1352.
+> **Claude Fable 5 (access via CC v2.1.170+)**: Mythos-class model, GA on the Claude API and positioned as a tier above Opus — its capabilities exceed any previously GA model. CC v2.1.170 is the client version that adds access (the model's GA is an API/platform property, not a CC-release milestone). Available via frontmatter full ID `claude-fable-5` (Tier 2) or Agent tool `model: fable` (Tier 3) — NOT via a Tier-1 frontmatter alias. Reserve for the most complex reasoning where its capability premium is warranted; `sonnet` remains the default for general tasks and `opus` for architecture (cost/latency awareness, R005). CC v2.1.170 also fixes session transcripts not saving (and not appearing in `--resume`) when launched from a VS Code integrated terminal or any shell inheriting Claude Code env vars — relevant to transcript-dependent skills (`homework`, `episodic-memory`). Closes #1352.
 
 <!-- ARCHIVED CC version notes (historical):
 > **v2.1.173+**: Fable 5 model IDs carrying a `[1m]` suffix are now auto-normalized (the suffix is stripped) because Fable 5 includes 1M context by default. Use `claude-fable-5` / `model: fable` WITHOUT a `[1m]` suffix — appending it is redundant and normalized away. (The `[1m]` suffix remains meaningful for Opus/Sonnet IDs.)
 
-> **v2.1.197+**: Claude Sonnet 5가 Claude Code의 **기본 모델**로 도입되었습니다 — 네이티브 1M-token 컨텍스트, 프로모션 가격 $2/$10 per Mtok(2026-08-31까지). `model: sonnet5` / `claude-sonnet-5`로 사용. oh-my-customcode의 base `sonnet` alias는 안정성을 위해 `claude-sonnet-4-6`에 고정 유지(기존 `sonnet` 지정 에이전트 불변); Sonnet 5는 `sonnet5`로 명시 opt-in. Sonnet 5가 CC 신규 기본값이므로 명시 모델 없는 세션은 이제 Sonnet 5에서 동작합니다.
+> **v2.1.197+**: Claude Sonnet 5가 Claude Code의 **기본 모델**로 도입되었습니다 — 네이티브 1M-token 컨텍스트, 프로모션 가격 $2/$10 per Mtok(2026-08-31까지). frontmatter에서 명시 opt-in하려면 Tier-2 full ID `claude-sonnet-5`를 사용합니다(`sonnet5`는 어느 계층에서도 유효한 값이 아님 — 위 3-Tier 구분 참조). **정정(실측)**: 이 조항이 이전에 "oh-my-customcode의 base `sonnet` alias는 안정성을 위해 `claude-sonnet-4-6`에 고정 유지"라고 서술했으나 사실이 아니다 — `sonnet` alias 해석 주체는 CC이며 프로젝트가 pin할 수 없다(Tier 1 참조); frontmatter `model: sonnet` 에이전트가 실측상 `claude-sonnet-5`로 실행되었다. Sonnet 5가 CC 신규 기본값이므로 명시 모델 없는 세션은 이제 Sonnet 5에서 동작합니다.
 
 > **v2.1.201+**: Claude Sonnet 5 세션이 harness reminder를 mid-conversation system role로 주입하지 않도록 변경되었습니다 — Sonnet 5 실행 시 하니스 리마인더(규칙 재주입 등) 전달 방식이 조정되었으며, PostCompact 규칙 재주입(R021)·세션 연속성 동작 자체에는 영향이 없습니다. Sonnet 5가 CC 기본 모델(v2.1.197+)이므로 명시 모델 없는 세션에 적용됩니다.
 -->
 
 > **Fable 5 Effort 전략**: Fable 5는 **high effort가 기본값**이며, `xhigh`는 capability-sensitive 작업(최고난도 아키텍처/추론)에 한정해야 합니다. Fable 5의 `low`/`medium` effort조차 이전 세대 모델의 `xhigh`를 상회하는 품질을 보이므로, Fable 5를 사용하는 실행 에이전트는 `effort` 필드를 신중히 명시하고 불필요한 `xhigh` 남용을 지양합니다(R005 비용/지연 인식과 정합).
 
-> **Mythos 5 (`claude-mythos-5`)**: Project Glasswing 한정 공급 모델로, **GA가 아닙니다** — Fable 5(GA, 위 Model Aliases 표의 `fable`)와 구분해야 합니다. 특성: adaptive-thinking 전용 아키텍처 + 안전 분류기가 개입 시 `stop_reason: "refusal"`로 fallback하는 체계를 가집니다. oh-my-customcode 에이전트 frontmatter에는 아직 alias를 등록하지 않습니다(비-GA, 공급 제한).
+> **Mythos 5 (`claude-mythos-5`)**: Project Glasswing 한정 공급 모델로, **GA가 아닙니다** — Fable 5(GA, 위 "Model Specification — 3 Tiers"의 `claude-fable-5`/`fable`)와 구분해야 합니다. 특성: adaptive-thinking 전용 아키텍처 + 안전 분류기가 개입 시 `stop_reason: "refusal"`로 fallback하는 체계를 가집니다. oh-my-customcode 에이전트 frontmatter에는 아직 alias를 등록하지 않습니다(비-GA, 공급 제한).
 
 > **프롬프팅 패턴 상호참조**: Fable 5/Mythos 5 대상 프롬프팅 패턴(effort 조합, adaptive-thinking 활용, refusal fallback 대응)의 상세 가이드는 `guides/claude-code/16-fable5-prompting.md`를 참조하세요.
 
@@ -67,7 +93,7 @@ This is a settings-level resilience mechanism, distinct from the per-agent `mode
 -->
 
 <!-- ARCHIVED CC version notes (historical):
-> **v2.1.183+**: CC now warns (stderr, in `-p` print mode) when the requested model is deprecated or auto-updated to a newer model — and this warning now ALSO covers models set in agent frontmatter (`model:`). Relevant to the Model Aliases table above: a stale/deprecated `model:` value in agent frontmatter now surfaces a deprecation warning instead of silently resolving. Separately, v2.1.183 fixes `thinking.disabled.display: Extra inputs are not permitted` 400 errors on subagent spawns and session-title generation — extends the v2.1.166 toggle above; subagent spawns with thinking disabled no longer 400.
+> **v2.1.183+**: CC now warns (stderr, in `-p` print mode) when the requested model is deprecated or auto-updated to a newer model — and this warning now ALSO covers models set in agent frontmatter (`model:`). Relevant to "Model Specification — 3 Tiers" above: a stale/deprecated `model:` value in agent frontmatter now surfaces a deprecation warning instead of silently resolving. Separately, v2.1.183 fixes `thinking.disabled.display: Extra inputs are not permitted` 400 errors on subagent spawns and session-title generation — extends the v2.1.166 toggle above; subagent spawns with thinking disabled no longer 400.
 
 > **v2.1.187+**: Org-configured model restrictions now apply to the model picker, `--model`, `/model`, and `ANTHROPIC_MODEL` (a restricted model shows "restricted by your organization's settings"). Extends the v2.1.175 `enforceAvailableModels` scope to per-agent `model:` override entry points — a managed model allowlist now also constrains the picker/env paths. Also fixed `--json-schema` / workflow `agent({schema})` structured output: the model can no longer re-call `StructuredOutput` indefinitely after a successful call, and follow-up turns reliably return structured output — relevant to schema-constrained subagent spawns.
 
@@ -110,7 +136,7 @@ source:                    # For external agents
   version: 1.0.0
 escalation:              # Model escalation policy (optional)
   enabled: true          # Enable auto-escalation advisory
-  path: haiku → sonnet → opus  # Escalation sequence
+  path: haiku → sonnet → opus  # Escalation sequence (Tier-3 Agent-tool aliases only)
   threshold: 2           # Failures before advisory
 soul: true                 # Enable SOUL.md identity injection
 isolation: worktree | sandbox  # worktree = git worktree, sandbox = restricted bash
@@ -297,7 +323,7 @@ When `escalation.enabled: true`, the model-escalation hooks will track outcomes
 | Field | Default | Description |
 |-------|---------|-------------|
 | `enabled` | false | Enable escalation tracking for this agent |
-| `path` | haiku → sonnet → opus | Model upgrade sequence |
+| `path` | haiku → sonnet → opus | Model upgrade sequence (Tier-3 Agent-tool aliases only) |
 | `threshold` | 2 | Failure count before escalation advisory |
 -->
 

package/templates/guides/agent-eval/README.md

@@ -824,5 +824,10 @@ harness-eval이 task를 실행하면 agent-eval-framework가 4-metric을 측정
 | 가이드 | 경로 | 관련성 |
 |--------|------|--------|
 | harness-engineering | `guides/harness-engineering/` | 평가 하네스 인프라 설계 |
-| monitoring-setup | `guides/monitoring-setup/` | OTEL 통합 (향후 tracing 보강) |
 | multi-agent-debate-patterns | `guides/multi-agent-debate-patterns/` | 다중 에이전트 평가 패턴 |
+
+### 관련 스킬
+
+| 스킬 | 경로 | 관련성 |
+|------|------|--------|
+| monitoring-setup | `.claude/skills/monitoring-setup/` | OTEL 통합 (향후 tracing 보강) |

package/templates/guides/claude-code/14-token-efficiency.md

@@ -224,7 +224,7 @@ External plugin (JuliusBrussee/caveman, 41.6k stars, MIT) that rewrites Claude r
 - `.claude/rules/SHOULD-ecomode.md` — Layer 2 R013 specification
 - `.claude/skills/token-efficiency-audit/SKILL.md` — Layer 3 HOW: audit and apply
 - `.claude/skills/monitoring-setup/SKILL.md` — Measure effectiveness via OTel metrics
-- `.claude/skills/update-config/` — Generic settings.json manipulation (broader scope)
+- CC bundled `update-config` skill — Generic settings.json manipulation (broader scope; not a project-local skill)
 - `.claude/skills/playwright-compress/SKILL.md` — Layer 4 MCP output compression hook
 - `.claude/hooks/scripts/playwright-compress.sh` — Layer 4 hook script
 - `https://github.com/JuliusBrussee/caveman` — Layer 5 caveman plugin (external)

package/templates/guides/claude-code/15-version-compatibility.md

@@ -3,11 +3,11 @@
 > Updated: 2026-07-02
 > Source: Claude Code release notes (#967, #968, #969, #1126 auto-detected by claude-native skill, #1137, #1158, #1242, #1243, #1244, #1245, #1276, #1280)
 >
-> **Note (compat 노트 이관)**: 이 파일은 v2.1.159까지의 호환성 노트를 보존합니다. v2.1.160+ 이후 CC 호환성 노트는 `.claude/rules/` 각 규칙(R001/R002/R006/R010/R012 등)에 인라인으로 축적(이관)되었으므로, 최신 compat 노트는 해당 규칙 파일을 참조하세요.
+> **Note (compat 노트 이관)**: 이 파일은 v2.1.160까지의 호환성 노트를 보존합니다. v2.1.161+ 이후 CC 호환성 노트는 `.claude/rules/` 각 규칙(R001/R002/R006/R010/R012 등)에 인라인으로 축적(이관)되었으므로, 최신 compat 노트는 해당 규칙 파일을 참조하세요.
 
 ## Compatibility Baseline
 
-oh-my-customcode v1.1.1 targets Claude Code v2.1.197+ (Sonnet 5 = CC 기본 모델). 이 파일에 정리된 v2.1.117-159 항목은 하위호환이며 config 변경이 불필요합니다.
+oh-my-customcode v1.1.9 targets Claude Code v2.1.201+ (Sonnet 5 = CC 기본 모델). 이 파일에 정리된 v2.1.117-160 항목은 하위호환이며 config 변경이 불필요합니다.
 
 ## v2.1.117 (2026-04-22)
 

package/templates/guides/deep-plan/phases.md

@@ -109,7 +109,7 @@ Phase 3: Plan Verification Research
 
 **Invocation**: Phase 3 teams are spawned directly as parallel agents (NOT via `Skill(research)`). The orchestrator creates 3 focused agents, each with a specific verification mandate derived from the Phase 2 plan.
 
-**Model selection**: sonnet for teams, opus for synthesis.
+**Model selection**: sonnet for teams, opus for synthesis (Agent-tool Tier 3 enum — see `guides/multi-model-routing/README.md`).
 
 ### Verdict
 

package/templates/guides/external-tools/ecc-absorption-decisions.md

@@ -78,6 +78,10 @@ R016 Continuous Improvement 루프를 강화하며, 내부 구조(agent-memory +
 삭제 경로가 불명확)를 profile 전환으로 해소합니다.
 cross-harness export 거부(#1176)로 발생하는 진입로 공백을 부분 보완합니다.
 
+> **후속 (v1.1.29, #1484)**: manifest profiles는 CLI 미배선(vestigial)으로 확정되어
+> 제거되었습니다. 이 흡수 결정 자체는 당시 문서화되었으나, 실제 CLI 배선(`--profile`
+> 플래그 실행 경로)은 이뤄지지 않았습니다.
+
 ---
 
 ## 3. 거부된 패턴 — Cross-harness Export (#1176)
@@ -136,7 +140,9 @@ multi-target 전환의 결과:
 | 기존 harness 자산 재활용 | 멀티 harness 사용자 |
 | ECC와 기능 중복 | zero-sum 경쟁 |
 
-신규 사용자 진입로 문제는 `manifest-install --profile`(#1177)으로 부분 해소됩니다.
+신규 사용자 진입로 문제는 `manifest-install --profile`(#1177)으로 부분 해소되는
+것으로 판단되었으나, 해당 profile 기능은 이후 CLI 미배선(vestigial)이 확인되어
+v1.1.29(#1484)에서 제거되었습니다.
 멀티 harness 사용자는 ECC를 직접 사용하는 것이 더 효과적입니다.
 
 ---
@@ -161,7 +167,8 @@ multi-target 전환의 결과:
 
 4. **진입로 문제 대체 해소**
    #1177 `manifest-install --profile`이 신규 사용자 진입 비용을 낮추는 대안으로
-   이미 흡수되었습니다.
+   당시 흡수되었습니다(단, v1.1.29/#1484에서 CLI 미배선 확인 후 제거됨 — 위
+   §2-3 후속 노트 참조).
 
 ---
 

package/templates/guides/harness-engineering/README.md

@@ -226,7 +226,7 @@ oh-my-customcode에서의 현황:
 # Break-out 시퀀스
 1. 현재 상태를 tracker-checkpoint에 기록
 2. [Warning] Doom loop detected — {action} × {n}회 반복
-3. 에스컬레이션: haiku → sonnet → opus 순서로 모델 전환
+3. 에스컬레이션: haiku → sonnet → opus 순서로 모델 전환 (Agent 도구 model 파라미터 — Tier 3 enum만 허용, sonnet5/opus5 등 전체 ID는 스폰 실패)
 4. 메모리 저장: feedback_doom_loop_{날짜}.md → 세션 간 패턴 누적
 ```
 

package/templates/guides/multi-model-routing/README.md

@@ -4,37 +4,62 @@
 
 Role-based model selection strategy for AI agent workflows. Consolidates model routing conventions from R006 (agent design), R008 (tool identification), and agent frontmatter into a single reference.
 
-## Model Aliases
+This project uses a **3-tier model specification system** — the same-looking string means a different thing depending on where it is placed. Read "Model Specification — 3 Tiers" below before choosing a value for frontmatter vs. an Agent tool spawn.
 
-| Alias | Full ID | Cost | Speed | Use Case |
-|-------|---------|------|-------|----------|
-| `haiku` | claude-haiku-4-5 | $ | Fast | Search, simple edits, file discovery |
-| `sonnet` | claude-sonnet-4-6 | $$ | Moderate | Code generation, general tasks (project base alias — pinned to 4-6 for stability) |
-| `sonnet5` | claude-sonnet-5 | $$ | Moderate | CC platform default (v2.1.197+); native 1M context |
-| `opus` | claude-opus-4-6 | $$$ | Slower | Complex reasoning, architecture, planning (project base alias — pinned to 4-6 for stability) |
-| `opus5` | claude-opus-5 | $$$ | Fast mode available | CC platform default Opus (v2.1.219+); native 1M context, fast mode at $10/$50 per Mtok |
-| `opus48` | claude-opus-4-8 | $$$ | Slower | Previous-generation Opus (GA); supports xhigh effort |
-| `fable` | claude-fable-5 | $$$$ | Slower | Mythos-class; tier above Opus, highest GA capability |
-| `opusplan` | claude-opus-4-6 + plan mode | $$$ | Slower | Architecture with approval gates |
+## Model Specification — 3 Tiers
 
-Extended context: `[1m]` suffix enables 1M token context (e.g., `claude-opus-4-6[1m]`). Fable 5, Sonnet 5, and Opus 5 include 1M context by default — no `[1m]` suffix needed.
+### Tier 1 — CC Native Alias
 
-> **Base alias pin (design intent)**: The base `sonnet` alias stays pinned to `claude-sonnet-4-6` for stability — existing agents specifying `model: sonnet` remain unchanged. Claude Sonnet 5 (`claude-sonnet-5`) is the CC platform default as of v2.1.197 and is opt-in via the explicit `sonnet5` alias. The same pattern applies to Opus: the base `opus` alias stays pinned to `claude-opus-4-6`; Claude Opus 5 (`claude-opus-5`) is CC's default Opus model as of v2.1.219 and is opt-in via the explicit `opus5` alias. Relative standing vs Fable 5 is not yet confirmed officially — do not assert an ordering.
+Values: `sonnet` | `opus` | `haiku` | `opusplan` | `inherit`
+
+Valid in BOTH agent frontmatter `model:` AND the Agent tool `model` parameter. **CC resolves these, not this project** — documentation claiming a pin (e.g. "the `sonnet` alias is pinned to 4-6") does not change what CC actually runs underneath. In practice, an agent with `model: sonnet` in frontmatter has been observed running as `claude-sonnet-5`.
+
+### Tier 2 — Full Model ID (frontmatter only)
+
+Values: `claude-sonnet-5` | `claude-opus-5` | `claude-haiku-4-5` | `claude-opus-4-6` | `claude-opus-4-8` | `claude-fable-5`, etc.
+
+Valid ONLY in frontmatter `model:` — the Agent tool `model` parameter does NOT accept full model IDs. Use a Tier 2 ID when you want the agent pinned to a specific model regardless of any future drift in what CC's Tier 1 aliases resolve to.
+
+### Tier 3 — Agent Tool `model` Parameter (enum)
+
+Values: `sonnet` | `opus` | `haiku` | `fable` — **only these four**. Does NOT accept full model IDs, and does NOT accept `sonnet5`/`opus5`-style names. Any spawn-time model escalation or routing table that dispatches via the Agent tool MUST use one of these four values.
+
+### Deprecated / Invalid Names
+
+`sonnet5`, `opus5`, and `opus48` are **not recognized by CC in any context** (neither frontmatter nor the Agent tool parameter) — they were project-invented names previously documented in this repo as aliases, but the platform does not interpret them; using them causes an immediate spawn/resolution failure. Bare `fable` (without the `claude-` prefix) is valid ONLY as a Tier 3 Agent tool parameter value — it is NOT a valid frontmatter alias; use the full ID `claude-fable-5` in frontmatter instead.
+
+## Model Catalog (Tier 2 full IDs)
+
+| Full Model ID | Cost | Speed | Use Case |
+|---------------|------|-------|----------|
+| `claude-haiku-4-5` | $ | Fast | Search, simple edits, file discovery |
+| `claude-sonnet-4-6` | $$ | Moderate | Code generation, general tasks — historically what the Tier 1 `sonnet` alias resolved to; CC's actual resolution target can drift (see Tier 1 above) |
+| `claude-sonnet-5` | $$ | Moderate | Native 1M context; use as an explicit frontmatter pin |
+| `claude-opus-4-6` | $$$ | Slower | Complex reasoning, architecture, planning |
+| `claude-opus-5` | $$$ | Fast mode available | Native 1M context, fast mode at $10/$50 per Mtok |
+| `claude-opus-4-8` | $$$ | Slower | Previous-generation Opus; supports xhigh effort |
+| `claude-fable-5` | $$$$ | Slower | Mythos-class; use the full ID in frontmatter (bare `fable` is Tier 3 Agent-tool-only) |
+
+Extended context: `[1m]` suffix enables 1M token context (e.g., `claude-opus-4-6[1m]`). Sonnet 5, Opus 5, and Fable 5 include 1M context by default — no `[1m]` suffix needed.
+
+> Relative standing of Fable 5 vs Opus 5 is not officially confirmed — do not assert an ordering.
 
 ## Role-Based Routing Table
 
-| Role | Recommended Model | Rationale |
-|------|------------------|-----------|
-| Code search / file discovery | haiku | Fast, cheap, sufficient for retrieval |
-| Code review | sonnet | Needs understanding, not deep reasoning |
-| Code generation | sonnet | Good balance of quality and speed |
-| Bug fix (simple) | sonnet | Pattern recognition sufficient |
-| Bug fix (complex) | opus | Needs deep reasoning across modules |
-| Architecture design | opus / opusplan | Requires holistic thinking |
-| Test generation | sonnet | Template-driven, moderate complexity |
-| Documentation | sonnet | Straightforward generation |
-| Release verification | opus | Cross-cutting validation |
-| Orchestration | opus | Routing decisions need broad context |
+This table recommends **frontmatter (Tier 2) full model IDs** for pinning an agent's default model. When dispatching the same role via the Agent tool `model` parameter (Tier 3) instead — e.g. a routing skill spawning a one-off agent — use the alias equivalent in the last column.
+
+| Role | Recommended Frontmatter ID (Tier 2) | Agent-Tool Equivalent (Tier 3) | Rationale |
+|------|--------------------------------------|----------------------------------|-----------|
+| Code search / file discovery | `claude-haiku-4-5` | `haiku` | Fast, cheap, sufficient for retrieval |
+| Code review | `claude-sonnet-5` | `sonnet` | Needs understanding, not deep reasoning |
+| Code generation | `claude-sonnet-5` | `sonnet` | Good balance of quality and speed |
+| Bug fix (simple) | `claude-sonnet-5` | `sonnet` | Pattern recognition sufficient |
+| Bug fix (complex) | `claude-opus-5` | `opus` | Needs deep reasoning across modules |
+| Architecture design | `claude-opus-5` / `opusplan` | `opus` | Requires holistic thinking |
+| Test generation | `claude-sonnet-5` | `sonnet` | Template-driven, moderate complexity |
+| Documentation | `claude-sonnet-5` | `sonnet` | Straightforward generation |
+| Release verification | `claude-opus-5` | `opus` | Cross-cutting validation |
+| Orchestration | `claude-opus-5` | `opus` | Routing decisions need broad context |
 
 ## Cost-Quality Tradeoff Matrix
 
@@ -56,16 +81,16 @@ Quality ▲
 
 ## MODEL_ROUTING.md Convention
 
-Projects can declare a `MODEL_ROUTING.md` file to override default routing:
+Projects can declare a `MODEL_ROUTING.md` file to override default routing. This convention sets a persistent per-agent frontmatter override, so use Tier 2 full model IDs:
 
 ```markdown
 # Model Routing
 
 | Agent Pattern | Model | Override Reason |
 |---------------|-------|-----------------|
-| lang-*-expert | sonnet | Default sufficient for code generation |
-| mgr-sauron | opus | Verification requires deep analysis |
-| Explore | haiku | Search-only, no generation needed |
+| lang-*-expert | claude-sonnet-5 | Default sufficient for code generation |
+| mgr-sauron | claude-opus-5 | Verification requires deep analysis |
+| Explore | claude-haiku-4-5 | Search-only, no generation needed |
 ```
 
 Place in project root or `.claude/` directory.
@@ -75,14 +100,14 @@ Place in project root or `.claude/` directory.
 ```yaml
 # .claude/agents/example.md
 name: example-agent
-model: sonnet  # Use alias from table above
+model: claude-sonnet-5  # Tier 2 full model ID (recommended) — see Model Specification — 3 Tiers above
 ```
 
-The `model` field in agent frontmatter sets the default. The Agent tool's `model` parameter overrides at spawn time.
+The `model` field in agent frontmatter accepts EITHER a Tier 1 alias (`sonnet`/`opus`/`haiku`/`opusplan`/`inherit`) OR a Tier 2 full model ID, and sets the agent's default. The Agent tool's `model` parameter at spawn time accepts ONLY the Tier 3 enum (`sonnet`/`opus`/`haiku`/`fable`) and overrides the frontmatter default when supplied.
 
 ## Escalation Pattern
 
-When a task fails at a lower model tier, escalate:
+When a task fails at a lower model tier, escalate. Escalation is executed by re-spawning via the Agent tool, so the path MUST use Tier 3 aliases — the Agent tool parameter does not accept full model IDs or `sonnet5`/`opus5`-style names:
 
 ```
 haiku → sonnet → opus
@@ -102,6 +127,6 @@ Fast Mode (`/fast` toggle) uses the same model with faster output (~2.5x). It do
 
 ## Related
 
-- R006 — Agent design rules (model aliases, frontmatter format)
+- R006 — Agent design rules (model tiers, frontmatter format)
 - R008 — Tool identification (model in agent:model format)
 - `guides/skill-bundle-design/` — Skill architecture patterns

package/templates/guides/multi-provider-exec/README.md

@@ -42,7 +42,7 @@ Providers are opt-in — missing CLIs are silently skipped.
 | Skill | Uses Provider | How |
 |-------|--------------|-----|
 | `reasoning-sandwich` | Any exec skill | Pre/post reasoning with different models |
-| `model-escalation` | Claude models only | Internal escalation (haiku→sonnet→opus), not cross-provider |
+| `model-escalation` | Claude models only | Internal escalation (haiku→sonnet→opus, Agent-tool Tier 3 enum), not cross-provider |
 
 ## Relationship to Multi-Model Routing
 

package/templates/manifest.json

@@ -1,5 +1,5 @@
 {
-  "version": "1.1.36",
+  "version": "1.1.38",
   "lastUpdated": "2026-07-14T00:00:00.000Z",
   "omcustomMinClaudeCode": "2.1.121",
   "omcustomMinClaudeCodeReason": "Sensitive-path direct Write/Edit on .claude/** under bypassPermissions (R010 deprecation, #1101)",

package/templates/workflows/auto-dev.yaml

@@ -230,15 +230,19 @@ steps:
 
       ## Local CI-mimic verification (MUST run before marking implement done)
 
-      After all implementation tasks complete, run these scripts in sequence and halt on failure:
+      After all implementation tasks complete, run these scripts in sequence and halt on failure.
+      Run each standalone (NO pipe) and read its exit code directly — `$?` after a pipe is the last command's code (R005, #1492):
 
       1. bash .github/scripts/verify-template-sync.sh
       2. bash .github/scripts/verify-wiki-sync.sh
+      3. bun .github/scripts/validate-docs.ts --programmatic-only
 
-      If either fails:
+      If any fails:
       - For wiki-sync failures: delegate to wiki-curator to generate missing pages (`/omcustom:wiki ingest <path>`)
+        (Regenerating the wiki source-hash manifest targets `wiki/.source-hashes.json` ALWAYS — NEVER `templates/manifest.json`. Command: `bash .github/scripts/lib/source-hash.sh generate wiki/.source-hashes.json`. Ref #1423.)
       - For template-sync failures: delegate to mgr-updater to sync `.claude/` -> `templates/.claude/`
-      - After fixes, re-run the scripts until both pass
+      - For validate-docs failures (count mismatch): delegate to mgr-updater to fix counts via EXHAUSTIVE grep across all docs — file-enumeration misses `.en`/`.ko` variants and secondary docs (#1521 — v1.1.32에서 validate-docs 누락으로 push 왕복 1회 발생)
+      - After fixes, re-run the scripts until all pass
 
       This mirrors CI and prevents the known 1-2 wasted push cycles per release (ref: issue #927, v0.98.0 PR #925).
     description: "Per-issue implementation with lifecycle, specialist routing, and CI-mimic verification"
@@ -299,10 +303,17 @@ steps:
          Determine NEW version per semver rules below.
          npm project (package.json exists):
            a. package.json: jq '.version = "<NEW>"' package.json > package.json.tmp && mv package.json.tmp package.json
+           # ⚠ jq '.version=...' PRESERVES the manifest structure {version, lastUpdated, omcustomMinClaudeCode, components[]}.
+           #   Edit ONLY the .version field — do NOT overwrite templates/manifest.json wholesale (e.g. a source-hash path→hash map).
+           #   Recover a corrupted manifest: git show HEAD:templates/manifest.json | jq '.version="<NEW>"' > templates/manifest.json (#1423/#1154).
            b. templates/manifest.json: jq '.version = "<NEW>"' templates/manifest.json > templates/manifest.json.tmp && mv templates/manifest.json.tmp templates/manifest.json
-           c. mgr-gitnerd commit: "chore(release): bump to v<NEW>"
-           d. mgr-gitnerd push develop
-           e. mandatory verification (with existence guard for partial-update safety):
+           c. bun run build — run standalone (NO pipe), read exit code directly ($? after a pipe is the last command's, R005/#1492).
+              This regenerates .omcustom.lock.json (generatorVersion/templateVersion) to <NEW> via scripts/sync-source-lockfile.ts.
+           d. git status --short — enumerate ALL tracked drift (`^ M`) the build produced; stage EVERY one (esp. .omcustom.lock.json) before commit.
+              (#1531 — .omcustom.lock.json missed staging across 4 consecutive releases after v1.1.29; root cause was no build step between version bump and commit.)
+           e. mgr-gitnerd commit: "chore(release): bump to v<NEW>" — MUST include the drift from step d
+           f. mgr-gitnerd push develop
+           g. mandatory verification (with existence guard for partial-update safety):
               [ -f scripts/verify-version-sync.sh ] && bash scripts/verify-version-sync.sh || echo "::warning::verify-version-sync.sh not found, version sync verification skipped"
               (verify-version-sync.sh 가 exit 1 시 release 단계 halt)
 
@@ -318,12 +329,15 @@ steps:
       3. Adapt release mechanism to project (determines how steps 3-5 execute):
          - npm project with auto-tag.yml (this repo — release/v* PR pattern):
            a. mgr-gitnerd creates release/v{NEW} branch from develop with the version-bump commit
-           b. mgr-gitnerd creates PR: gh pr create --base develop --head release/v{NEW} --title "chore(release): bump to v{NEW}"
+           b. mgr-gitnerd creates PR: gh pr create --base develop --head release/v{NEW} --title "chore(release): bump to v{NEW}" --body "<MUST include 'Closes #N' for EVERY issue this release resolves>"
+              ⚠ auto-tag.yml closes issues by grep'ing Closes/Fixes/Resolves keywords in the PR BODY — NOT by milestone membership. If the keywords are missing, no issues close and the workflow still reports success (#1531).
+              Before creating the PR (whether via inline --body or --body-file), confirm the body text actually contains a Closes line per issue.
            c. mgr-gitnerd merges PR (admin merge required due to branch protection)
            d. DO NOT manually git tag or gh release create.
-              auto-tag.yml fires on release/v* PR merge → creates tag → closes linked issues → closes milestone → deletes release branch.
+              auto-tag.yml fires on release/v* PR merge → creates tag → closes issues linked via PR-body Closes/Fixes/Resolves keywords → closes milestone → deletes release branch.
               release.yml fires on the tag → npm publish + GitHub Release creation.
            Milestone close and issue close are handled downstream by auto-tag.yml — do NOT close them manually.
+           After merge, verify each targeted issue is actually CLOSED (gh issue view); close manually if still open. Workflow conclusion=success is not evidence of close (#1531 — v1.1.34 Closes-keyword omission left 5 issues open, manual close required).
          - Non-npm / no auto-tag.yml:
            mgr-gitnerd: git tag v{NEW} && git push origin v{NEW}
            mgr-gitnerd: gh release create v{NEW} (with release notes)