docs: update all documentation for v0.1.43

- CHANGELOG: add v0.1.43 entry with 7 BMAD patterns + write-file + checklist
- README/README_CN: bump badge to v0.1.43, fix command count (37→39), add What's New section
- docs/flowctl.md: add write-file and checklist command documentation
- docs/skills.md: update skill count (26→47 extensions, 54 total), add Architecture & Infrastructure category
- CLAUDE.md: add zero-findings-halt rule to Quality Gates section
- docs/comparison: fix command count references

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: z23cc

CHANGELOG.md

@@ -4,6 +4,27 @@ All notable changes to Flow-Code are documented in this file.
 
 Format follows [Keep a Changelog](https://keepachangelog.com/). Versions use [Semantic Versioning](https://semver.org/).
 
+## [0.1.43] - 2026-04-08
+
+### Added
+- **7 BMAD-METHOD patterns** adopted from deep analysis of BMAD-METHOD v6.2.2
+- `flowctl write-file` — Pipeline file I/O command (bypasses Claude Code permission prompts for zero-interaction pipelines)
+- `flowctl checklist` — Structured Definition of Done with init/check/uncheck/verify/show subcommands (8 default items across 4 categories)
+- `project-context.md` support — Shared technical standards document auto-loaded by Worker Phase 2 re-anchoring (template in `templates/`)
+- Zero-findings-halt review rule — Reviewers must find issues; zero findings triggers NEEDS_REANALYSIS
+- Three-layer parallel code review — Blind Hunter (diff-only) + Edge Case Hunter (boundary analysis) + Acceptance Auditor (spec compliance)
+- Advanced elicitation methods in brainstorm — Pre-mortem, First Principles, Inversion, Red Team, Constraint Removal, Stakeholder Mapping
+- `--quick` flag for `/flow-code:go` — Fast path skipping brainstorm/plan for trivial changes
+- Step-file architecture — 15 step files across plan (5), work (5), brainstorm (5) for JIT loading
+- 3 deep comparison documents: flow-code vs compound-engineering, flow-code vs BMAD-METHOD, BMAD lessons analysis
+
+### Changed
+- Skills now use step-file workflow (JIT loading) instead of monolithic steps.md/phases.md
+- Removed legacy workflow files (steps.md, phases.md, examples.md) in favor of steps/ directories
+- `flowctl parse-findings` now detects zero findings and returns NEEDS_REANALYSIS verdict
+- Worker agent Phase 2 now reads `.flow/project-context.md` if present
+- `flowctl init` now hints about project-context.md template
+
 ## [0.1.42] - 2026-04-08
 
 ### Added

CLAUDE.md

@@ -45,7 +45,7 @@ Every epic passes through three independent, non-overlapping review gates:
 | **2. RP Plan-Review** | RP context_builder or Codex | Plan phase | Spec-code misalignment, missing requirements |
 | **3. Codex Adversarial** | `flowctl codex adversarial` | Epic completion | Security, concurrency, edge cases (different model family) |
 
-All three must pass (or be skipped via `flowctl config set review.backend none`). Layers are complementary — guard catches syntax, RP catches spec drift, Codex catches blind spots.
+All three must pass (or be skipped via `flowctl config set review.backend none`). Layers are complementary — guard catches syntax, RP catches spec drift, Codex catches blind spots. **Zero-findings-halt rule**: review cycles stop immediately when no findings remain, eliminating unnecessary re-review iterations. Conversely, zero findings from an adversarial review triggers re-analysis (review may be insufficient).
 
 ## Command Flags
 

README.md

@@ -7,7 +7,7 @@
 [![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](../../LICENSE)
 [![Claude Code](https://img.shields.io/badge/Claude_Code-Plugin-blueviolet)](https://claude.ai/code)
 
-[![Version](https://img.shields.io/badge/Version-0.1.42-green)](https://github.com/z23cc/flow-code/releases)
+[![Version](https://img.shields.io/badge/Version-0.1.43-green)](https://github.com/z23cc/flow-code/releases)
 
 [![Status](https://img.shields.io/badge/Status-Active_Development-brightgreen)](https://github.com/z23cc/flow-code/releases)
 
@@ -29,7 +29,7 @@ A **harness** wraps around an AI coding agent to handle everything the model can
 
 | Capability | Flow-Code | [compound-engineering](https://github.com/EveryInc/compound-engineering-plugin) (12.5K⭐) | [claude-mem](https://github.com/thedotmack/claude-mem) (44K⭐) | [superpowers](https://github.com/anthropics/claude-plugins-official) |
 |---|---|---|---|---|
-| Task DAG + state machine | ✅ 37 commands, deps, split/skip | ❌ | ❌ | ❌ |
+| Task DAG + state machine | ✅ 39 commands, deps, split/skip | ❌ | ❌ | ❌ |
 | Parallel Teams + file lock | ✅ Agent Teams, atomic locks | ❌ | ❌ | ✅ parallel agents (no locks) |
 | Three-layer quality | ✅ guard + RP + Codex adversarial | ❌ | ❌ | ❌ |
 | Runtime DAG mutation | ✅ split/skip/dep rm mid-execution | ❌ | ❌ | ❌ |
@@ -39,6 +39,17 @@ A **harness** wraps around an AI coding agent to handle everything the model can
 | Auto draft PR | ✅ | ❌ | ❌ | ❌ |
 | Minimal dependencies | ✅ single Rust binary + git/jq/gh | ❌ Node.js | ❌ ChromaDB | ❌ Node.js |
 
+### What's New in v0.1.43
+
+- **`write-file` command** — pipeline helper that writes content to files, bypassing permission prompts
+- **`checklist` command** — structured Definition of Done checklists for epics
+- **`project-context.md`** — auto-generated project context document for onboarding and agent orientation
+- **Step-file architecture** — modular step files replace monolithic skill definitions, enabling per-step overrides and composition
+- **3-layer parallel code review** — guard, RP plan-review, and Codex adversarial now run concurrently where possible
+- **Zero-findings-halt rule** — review cycles stop immediately when no findings remain, eliminating unnecessary re-review iterations
+- **Quick-dev fast path** — small, well-scoped changes skip full planning overhead and execute directly
+- **Advanced elicitation** — improved requirement extraction via structured probing during brainstorm and interview phases
+
 ---
 
 ## Table of Contents

README_CN.md

@@ -7,7 +7,7 @@
 [![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](../../LICENSE)
 [![Claude Code](https://img.shields.io/badge/Claude_Code-Plugin-blueviolet)](https://claude.ai/code)
 
-[![Version](https://img.shields.io/badge/Version-0.1.42-green)](../../CHANGELOG.md)
+[![Version](https://img.shields.io/badge/Version-0.1.43-green)](../../CHANGELOG.md)
 
 [![Status](https://img.shields.io/badge/Status-Active_Development-brightgreen)](../../CHANGELOG.md)
 
@@ -29,14 +29,25 @@
 
 | 能力 | Flow-Code | [compound-engineering](https://github.com/EveryInc/compound-engineering-plugin) (12.5K⭐) | [claude-mem](https://github.com/thedotmack/claude-mem) (44K⭐) | [superpowers](https://github.com/anthropics/claude-plugins-official) |
 |---|---|---|---|---|
-| Task DAG + 状态机 | ✅ 37 命令，依赖图，split/skip | ❌ | ❌ | ❌ |
+| Task DAG + 状态机 | ✅ 39 命令，依赖图，split/skip | ❌ | ❌ | ❌ |
 | Teams 并行 + 文件锁 | ✅ Agent Teams，原子锁 | ❌ | ❌ | ✅ 并行 agents（无锁） |
 | 三层质量体系 | ✅ guard + RP + Codex 对抗 | ❌ | ❌ | ❌ |
 | 运行时 DAG 变更 | ✅ 执行中 split/skip/dep rm | ❌ | ❌ | ❌ |
 | 跨模型对抗审查 | ✅ GPT 试图破坏 Claude 的代码 | ❌ | ❌ | ❌ |
 | 全自动（零问题） | ✅ AI 决定 branch/review/depth | ❌ | ❌ | ❌ |
 | 零依赖 | ✅ 单 Rust 二进制 + Bash 技能 | ❌ Node.js | ❌ ChromaDB | ❌ Node.js |
 
+### v0.1.43 新增功能
+
+- **`write-file` 命令** — 流水线辅助工具，写入文件内容，绕过权限提示
+- **`checklist` 命令** — 为 epic 提供结构化的完成定义检查清单
+- **`project-context.md`** — 自动生成的项目上下文文档，用于入职和 agent 定向
+- **Step-file 架构** — 模块化步骤文件取代单体 skill 定义，支持逐步覆盖和组合
+- **三层并行代码审查** — guard、RP plan-review 和 Codex 对抗审查现在可以并行运行
+- **零发现即停止规则** — 当没有剩余发现时，审查循环立即停止，消除不必要的重复审查
+- **Quick-dev 快速路径** — 小型、范围明确的变更跳过完整规划开销，直接执行
+- **高级需求引导** — 在头脑风暴和访谈阶段通过结构化探针改进需求提取
+
 ---
 
 ## 目录

docs/comparison-flow-code-vs-compound.md

@@ -157,7 +157,7 @@ compound-engineering's multi-platform conversion system is a significant differe
 
 ### Analysis
 
-flow-code provides richer CLI tooling (37 commands with JSON output), more autonomous operation by default, and specialized tools (debug, map, prime, auto-improve) that compound lacks. compound-engineering provides better interactive workflows, universal planning for non-software tasks, and structured ideation.
+flow-code provides richer CLI tooling (39 commands with JSON output), more autonomous operation by default, and specialized tools (debug, map, prime, auto-improve) that compound lacks. compound-engineering provides better interactive workflows, universal planning for non-software tasks, and structured ideation.
 
 ---
 
@@ -245,7 +245,7 @@ flow-code provides richer CLI tooling (37 commands with JSON output), more auton
 | **Quality** | 3 layers + circuit breaker | 20+ personas + autofix routing |
 | **Knowledge** | Auto-capture + staleness check | Compounding + dual-track schemas |
 | **Platforms** | Claude + Codex sync | 10+ platform targets |
-| **CLI** | 37 commands with --json | No CLI (converter only) |
+| **CLI** | 39 commands with --json | No CLI (converter only) |
 | **Safety** | File locking + invariants + gaps | Convention-based |
 | **Learning** | Retro + memory + auto-improve | Compound + refresh |
 | **Rollout** | No beta framework | Beta skills + safe rollouts |

docs/flowctl.md

@@ -1317,6 +1317,42 @@ flowctl import [--json]
 
 Rebuilds the DB from `.flow/` Markdown files.
 
+### write-file
+
+Write content to a file. Pipeline helper that bypasses Claude Code permission prompts.
+
+```bash
+# Inline content
+flowctl write-file --path "path/to/file.md" --content "content here" --json
+
+# Stdin (for long content via heredoc)
+cat <<'EOF' | flowctl write-file --path "path/to/file.md" --stdin --json
+Long content here...
+EOF
+
+# Append mode
+flowctl write-file --path "path/to/file.md" --content "new line" --append --json
+```
+
+Options:
+- `--path` (required) — Target file path. Creates parent directories if needed.
+- `--content` — Inline content string
+- `--stdin` — Read content from stdin
+- `--append` — Append instead of overwrite
+
+### checklist
+
+Structured Definition of Done checklists for tasks. 8 default items across 4 categories (context, implementation, testing, documentation).
+
+Subcommands:
+- `checklist init --task <id>` — Create default DoD checklist for a task
+- `checklist check --task <id> --item <key>` — Mark item as checked
+- `checklist uncheck --task <id> --item <key>` — Unmark item
+- `checklist verify --task <id>` — Verify all items; exits 1 if any missing
+- `checklist show --task <id>` — Display current checklist state
+
+Storage: `.flow/checklists/<task-id>.json`
+
 ### completions
 
 Generate shell completions.

docs/skills.md

@@ -16,17 +16,18 @@ These skills form the primary workflow. They ship with the plugin.
 | `flow-code-setup` | `/flow-code:setup` | Install flowctl CLI and configure project | `skills/` |
 | `flow-code-map` | `/flow-code:map` | Generate codebase architecture maps | `skills/` |
 
-## Extension Skills (26)
+## Extension Skills (47)
 
-Optional capabilities that extend the core workflow.
+Optional capabilities that extend the core workflow. Skills live in `skills/` and/or `codex/skills/` (54 unique skills total, 7 core + 47 extension).
 
-### Review & Quality (codex/skills/)
+### Review & Quality
 
 | Skill | Command | Purpose |
 |-------|---------|---------|
 | `flow-code-plan-review` | Internal | Cross-model plan validation (RP or Codex) |
 | `flow-code-impl-review` | Internal | Per-task implementation review with fix loop |
 | `flow-code-epic-review` | Internal | Epic completion adversarial gate |
+| `flow-code-code-review` | Internal | Five-axis scoring with severity labels (Worker Phase 6, impl-review, PR review) |
 
 ### Development Extensions
 
@@ -39,6 +40,28 @@ Optional capabilities that extend the core workflow.
 | `flow-code-deps` | `/flow-code:deps` | Dependency graph visualization and execution order |
 | `flow-code-api-design` | `/flow-code:api-design` | API design and module boundary review |
 | `flow-code-performance` | `/flow-code:performance` | Performance investigation, optimization, and benchmarks |
+| `flow-code-tdd` | `/flow-code:tdd` | Test-first methodology, Prove-It Pattern, coverage strategy |
+| `flow-code-incremental` | Internal | Vertical slicing, incremental commits, Implement-Test-Verify-Commit cycle |
+| `flow-code-simplify` | `/flow-code:simplify` | Reduce code complexity while preserving exact behavior |
+| `flow-code-frontend-ui` | `/flow-code:frontend-ui` | Production-quality UI: components, layouts, state, accessibility |
+
+### Architecture & Infrastructure Skills
+
+| Skill | Command | Purpose |
+|-------|---------|---------|
+| `flow-code-security` | Internal | OWASP Top 10, three-tier security boundaries |
+| `flow-code-auth` | Internal | Authentication, authorization, OAuth, JWT, RBAC, session management |
+| `flow-code-database` | Internal | Schema design, migrations, query optimization, ORM patterns |
+| `flow-code-caching` | Internal | HTTP cache, CDN, Redis, in-memory, cache invalidation strategy |
+| `flow-code-containerization` | Internal | Dockerfiles, docker-compose, Kubernetes, image optimization |
+| `flow-code-microservices` | Internal | Service boundaries, inter-service communication, saga, event-driven |
+| `flow-code-realtime` | Internal | WebSocket, SSE, long-polling, connection management, scaling |
+| `flow-code-state-management` | Internal | Frontend/full-stack state architecture, tool selection, patterns |
+| `flow-code-error-handling` | Internal | Error classification, retry logic, circuit breakers, graceful degradation |
+| `flow-code-i18n` | Internal | Multi-language support, locale formatting, RTL layouts |
+| `flow-code-monitoring` | Internal | Dashboards, SLOs/SLIs, alerting rules, on-call runbooks |
+| `flow-code-observability` | Internal | Logging, tracing, metrics, health endpoints |
+| `flow-code-documentation` | Internal | ADRs, API docs, READMEs, changelogs, doc-as-code workflow |
 
 ### Workflow Extensions
 
@@ -51,6 +74,7 @@ Optional capabilities that extend the core workflow.
 | `flow-code-autoplan` | `/flow-code:autoplan` | Multi-perspective auto-review pipeline (CEO, eng, design, DX) |
 | `flow-code-qa` | `/flow-code:qa` | Visual QA testing with browser automation |
 | `flow-code-design-review` | `/flow-code:design-review` | Visual design audit with browser automation |
+| `flow-code-guide` | Internal | Skill discovery flowchart — helps select the right skill |
 
 ### Tooling Extensions
 
@@ -99,6 +123,7 @@ Simple utilities and wrappers that perform a single focused action.
 | `flow-code-export-context` | Export context for external model review |
 | `flow-code-deps` | Dependency graph visualization |
 | `flow-code-loop-status` | Monitor running loops |
+| `flow-code-guide` | Skill discovery flowchart |
 | `browser` | Browser automation via agent-browser CLI |
 | `flow-code-rp-explorer` | RepoPrompt-powered codebase exploration |
 
@@ -114,6 +139,9 @@ Skills that gather information, analyze codebases, or provide monitoring and gui
 | `flow-code-performance` | Performance investigation and benchmarks |
 | `flow-code-prompt-eng` | Prompt engineering guidance for review agents |
 | `flow-code-debug` | Systematic debugging with root cause investigation |
+| `flow-code-monitoring` | Dashboards, SLOs/SLIs, alerting rules |
+| `flow-code-observability` | Logging, tracing, metrics, health endpoints |
+| `flow-code-documentation` | ADRs, API docs, changelogs, doc-as-code |
 
 ### Tier 3 — Interactive Decision-Making
 
@@ -138,9 +166,24 @@ Skills that involve planning, design, or interactive multi-step workflows.
 | `flow-code-plan-review` | Cross-model plan validation |
 | `flow-code-impl-review` | Per-task implementation review |
 | `flow-code-epic-review` | Epic completion adversarial gate |
+| `flow-code-code-review` | Five-axis scoring with severity labels |
 | `flow-code-qa` | Visual QA testing with browser |
 | `flow-code-design-review` | Visual design audit with browser |
 | `flow-code-autoplan` | Multi-perspective auto-review pipeline |
+| `flow-code-tdd` | Test-first methodology, Prove-It Pattern |
+| `flow-code-incremental` | Vertical slicing, incremental commits |
+| `flow-code-simplify` | Reduce complexity while preserving behavior |
+| `flow-code-frontend-ui` | Production-quality UI components and layouts |
+| `flow-code-security` | OWASP Top 10, security boundaries |
+| `flow-code-auth` | Authentication, authorization, OAuth, JWT, RBAC |
+| `flow-code-database` | Schema design, migrations, query optimization |
+| `flow-code-caching` | HTTP cache, CDN, Redis, cache invalidation |
+| `flow-code-containerization` | Docker, Kubernetes, image optimization |
+| `flow-code-microservices` | Service boundaries, saga, event-driven patterns |
+| `flow-code-realtime` | WebSocket, SSE, connection management |
+| `flow-code-state-management` | Frontend/full-stack state architecture |
+| `flow-code-error-handling` | Error classification, retry, circuit breakers |
+| `flow-code-i18n` | Multi-language support, locale formatting, RTL |
 
 ### Tier 4 — Final Decisions & Execution