release: v2.9.1

[GoosLab]

Release v2.9.1

Changes

• Fix: Agency v3.2 template deployment (35 files missing from moai update)
• CLAUDE.md v14.0.0: Agency command reference, agent catalog, harness routing
• README: Agency v3.2 section in 4 languages (en, ko, ja, zh)
• CLAUDE.local.md: Template-First Rule added

Quality Gates

• Tests: PASS (38 packages, race detection)
• go vet: PASS
• gofmt: CLEAN

Auto-generated by MoAI release workflow.

Summary by CodeRabbit

Release Notes v2.9.1

• New Features

  • Introduced AI Agency v3.2 system with unified /agency command for creative production workflows
  • Added new Agency agents: planner, copywriter, designer, builder, evaluator, and learner with specialized roles
  • Implemented harness-based quality evaluation with iterative feedback loops and multi-dimensional scoring
  • Added agency subcommands: brief, build, review, learn, evolve, resume, profile, sync-upstream, rollback, config, and phase

• Documentation

  • Updated documentation with AI Agency v3.2 specifications and agent catalog
  • Added agency configuration and context sections

[coderabbitai]

Caution

Review failed

Pull request was closed or merged during review

Walkthrough

This PR bumps MoAI version to v2.9.1 and introduces AI Agency v3.2—a comprehensive multi-agent creative production framework. It adds 35+ template files defining agent roles, pipeline orchestration, configuration schemas, context templates, and constitutional rules. The release documents the template inclusion and version updates across configuration and codebase.

Changes

Cohort / File(s)	Summary
Version Bump .moai/config/sections/system.yaml, pkg/version/version.go	Updated moai.template_version and moai.version from v2.9.0 to v2.9.1; updated Go package Version constant from v2.9.0 to v2.9.1.
Release Documentation CHANGELOG.md	Added v2.9.1 patch release entry documenting Agency v3.2 template deployment fix, template file inclusion, CLAUDE.md documentation bump to v14.0.0, and README updates across multiple languages.
Local Development Guidelines CLAUDE.local.md	Expanded "Template Source" section to include .agency/ directory; added hard rule requiring template-first workflow for new files; extended "Local-Only Files" list with runtime-generated items and refined entry descriptions.
Template Documentation CLAUDE.md, internal/template/templates/CLAUDE.md	Added /agency unified skill with pipeline routing (Planner → Copywriter/Designer → Builder → Evaluator → Learner) and GAN-loop iteration; documented six Agency agents and evaluator-active variant; introduced Harness-Based Quality Routing with Complexity Estimator; added Agency rules and configuration references; bumped template version to 14.0.0.
Agency Core Configuration internal/template/templates/.agency/config.yaml	Added AI Agency v3.2 top-level configuration with pipeline phases, GAN-loop settings (iterations, thresholds, escalation), evolution parameters (auto-thresholds, learning caps, approval requirements), context/brand settings, model assignments per agent role, pipeline adaptation controls, user profile, and integration toggles.
Agency Context Templates internal/template/templates/.agency/context/{brand-voice,quality-standards,target-audience,tech-preferences,visual-identity}.md	Added five Markdown context templates defining Brand Voice (identity, tone spectrum, language rules, example phrases), Quality Standards (performance/accessibility/SEO/browser/code/testing/security requirements), Target Audience (personas, customer journey, decision factors, objections, channels), Tech Preferences (frontend/styling/component/language/build/hosting/integrations/dev-env), and Visual Identity (colors, typography, logo, imagery, layout, anti-patterns).
Agency Fork Manifest & Brief Template internal/template/templates/.agency/fork-manifest.yaml, internal/template/templates/.agency/templates/brief-template.md	Added fork manifest (v2 schema) tracking agent/skill forks with upstream refs and sync policies; added brief template with project metadata, business/brand/content/tech sections, evaluation criteria, and GAN-loop progress appendix.
Agency Agent Definitions internal/template/templates/.claude/agents/agency/{planner,copywriter,designer,builder,evaluator,learner}.md	Added six agent configuration files defining role-specific identity, frozen safety rails, ethical boundaries, and evolvable operational constraints; Planner handles client interviews and brief generation; Copywriter produces JSON-structured page content; Designer creates design specifications; Builder implements TDD-based production code; Evaluator performs Playwright-based quality scoring with hard-fail thresholds; Learner orchestrates evolution pipeline and upstream sync.
Agency Command Routers internal/template/templates/.claude/commands/agency/{agency,brief,build,review,learn,evolve,resume,profile}.md	Added command definition files (one master orchestrator plus subcommand routers) specifying skill invocation patterns and argument hints for the agency workflow (brief, build, review, learn, evolve, resume, profile, sync-upstream, rollback, config).
Agency Constitution & Rules internal/template/templates/.claude/rules/agency/constitution.md	Added comprehensive constitution (445 lines) defining immutable vs. evolvable zones, brand-context enforcement, pipeline contracts, five-layer safety architecture for evolution (frozen guard, canary evaluation, contradiction detection, rate limiting, human oversight), learnings pipeline with anti-pattern detection, fork/sync rules with manifest requirements, GAN-loop contract (pass/fail thresholds, escalation, stagnation detection), and error recovery behaviors.
Agency Skills internal/template/templates/.claude/skills/agency{,-client-interview,-copywriting,-design-system,-evaluation-criteria,-frontend-patterns}/SKILL.md	Added six skill definition files: agency (orchestrator with subcommand routing and pipeline execution); agency-client-interview (progressive-disclosure interview framework); agency-copywriting (JSON output contracts for page sections); agency-design-system (visual design governance); agency-evaluation-criteria (Playwright testing, skeptical scoring, hard-fail rules); agency-frontend-patterns (TDD, accessibility, performance budgets).

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Possibly related PRs

• #602 — Implements the same /agency skill router and subcommand wrapper pattern for orchestrating the multi-agent pipeline.
• #601 — Introduces AI Agency v3.2 template structure with overlapping agent, skill, command, and configuration files.
• #519 — Updates version values in .moai/config/sections/system.yaml and package version constants.

Suggested labels

docs, config, go

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'release: v2.9.1' directly and clearly describes the main change: a version release bump. It follows conventional commit format, is concise, and accurately reflects the primary objective.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch release/v2.9.1

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[GoosLab]

@claude CI 빌드가 실패했습니다. 오류를 분석하고 수정해주세요.

실패한 워크플로우: Claude Code Review
브랜치: release/v2.9.1
실패 로그: https://github.com/modu-ai/moai-adk/actions/runs/23930506488

실패 로그를 확인하고 근본 원인을 파악한 뒤 수정해주세요.

Latte AI CI Monitor • latte@mo.ai.kr