feat: implement 7 BMAD-METHOD patterns + write-file command (v0.1.43)

New features inspired by BMAD-METHOD deep analysis:
1. project-context.md shared context — Worker Phase 2 reads .flow/project-context.md
2. Zero-findings-halt review rule — reviewers must find issues, zero = reanalysis
3. Structured DoD checklist — flowctl checklist init/check/verify (8 items, 4 categories)
4. Quick-dev fast path — --quick flag skips brainstorm/plan for trivial changes
5. Advanced elicitation — Pre-mortem, First Principles, Inversion in brainstorm
6. Step-file architecture — 15 step files across plan/work/brainstorm (JIT loading)
7. 3-layer parallel code review — Blind Hunter + Edge Case Hunter + Acceptance Auditor

Also adds flowctl write-file command for zero-interaction pipeline file I/O,
and 3 deep comparison documents (vs compound-engineering, vs BMAD-METHOD).

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

Author: z23cc

.claude-plugin/marketplace.json

@@ -6,13 +6,13 @@
   },
   "metadata": {
     "description": "Structured plan-first development plugin for Claude Code with .flow/ task tracking, Agent Teams, and Ralph autonomous mode",
-    "version": "0.1.42"
+    "version": "0.1.43"
   },
   "plugins": [
     {
       "name": "flow-code",
       "description": "Zero-dependency planning + execution with .flow/ task tracking. Three-layer quality (guard + RP plan-review + Codex adversarial). Full-auto, zero questions. Teams auto-parallel, DAG mutation, Codex-driven decisions, auto draft-PR, session summary. 5 prompt templates. 24 subagents, 22 commands, 74 skills.",
-      "version": "0.1.42",
+      "version": "0.1.43",
       "source": "./",
       "category": "workflow",
       "tags": [
@@ -26,5 +26,5 @@
       "strict": true
     }
   ],
-  "version": "0.1.42"
+  "version": "0.1.43"
 }

.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "flow-code",
-  "version": "0.1.42",
+  "version": "0.1.43",
   "description": "Zero-dependency planning + execution with .flow/ task tracking and Ralph autonomous mode (multi-model review gates). Worker subagent per task with git worktree isolation for parallel execution. Three-layer quality system (guard + RP plan-review + Codex adversarial). Full-auto by default — AI decides from context, zero questions. Teams-default with file locking, DAG mutation, Codex-driven conflict resolution, auto draft-PR. Auto-detected stack profiles with one-command guard (test/lint/typecheck). Enhanced agent definitions with permissionMode/maxTurns/effort. Lifecycle hooks with state preservation (PreCompact injects .flow state into compaction, TaskCompleted auto-unlocks files, SubagentStart context injection). Memory v2 with atomic entries, dedup, and progressive disclosure. TDD enforcement mode. Multi-epic queue with dependency visualization. Includes 24 subagents, 22 commands, 74 skills.",
   "author": {
     "name": "z23cc",

.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "flow-code",
-  "version": "0.1.42",
+  "version": "0.1.43",
   "description": "Zero-dependency planning + execution with .flow/ task tracking and Ralph autonomous mode (multi-model review gates). Worker subagent per task with git worktree isolation for parallel execution. Three-layer quality system (guard + RP plan-review + Codex adversarial). Full-auto by default. Teams-default with file locking, DAG mutation, Codex-driven conflict resolution, auto draft-PR.",
   "author": {
     "name": "z23cc",

CLAUDE.md

@@ -60,6 +60,7 @@ All three must pass (or be skipped via `flowctl config set review.backend none`)
 | `--research=rp\|grep` | plan | Override research backend |
 | `--depth=short\|standard\|deep` | plan | Override plan depth |
 | `--review=rp\|codex\|none` | plan, run | Override review backend |
+| `--quick` | go, run | Fast path for trivial changes (skip brainstorm, plan review, impl review) |
 
 ## Worker Protocol (Teams Mode)
 
@@ -149,6 +150,7 @@ Rust: clippy for linting, cargo test for tests. No TypeScript, no npm. Skills an
 - **Worker timeout**: Workers have a 30-minute default timeout per task (configurable via `worker.timeout_minutes`). On timeout: task marked failed, file locks released, wave continues
 - **Stale lock recovery**: Runs at wave start AND on worker completion — detects locks held by done/failed/blocked tasks and releases them to prevent deadlocks
 - **Worker phase mapping**: Workers execute 12 internal phases (via `flowctl worker-phase next/done`) within the epic "Work" phase. Epic phases and worker phases are independent systems operating at different levels
+- **Project context**: Optional `.flow/project-context.md` (template in `templates/project-context.md`) provides shared technical standards (stack, rules, architecture decisions, non-goals) that all worker agents read during Phase 2 re-anchoring. Keeps agents aligned on conventions code alone can't convey
 
 ## Files to Never Commit
 

agents/worker.md

@@ -209,6 +209,13 @@ git log -5 --oneline
 <FLOWCTL> config get memory.enabled --json
 ```
 
+**Read project context** (if `.flow/project-context.md` exists):
+```bash
+# Load shared technical standards (non-negotiable rules, stack details, architecture decisions)
+cat .flow/project-context.md 2>/dev/null
+```
+If the file exists, treat its contents as authoritative project-wide constraints. Apply any rules from "Critical Implementation Rules" and "Non-Goals" throughout all subsequent phases.
+
 **If memory.enabled is true**, inject relevant memory (L1: compact index):
 ```bash
 <FLOWCTL> memory inject --json

bin/flowctl

@@ -94,6 +94,33 @@ Wait for each answer before asking the next question.
 
 After all 3 answers, summarize the key insights in 2-3 bullets before proceeding.
 
+## Structured Deepening
+
+After self-interview completes, apply 1-2 named reasoning methods to pressure-test the output:
+
+### Method Selection Guide
+| Method | Best for | Prompt |
+|--------|----------|--------|
+| **Pre-mortem Analysis** | Specs, plans, new features | "Assume this shipped and failed 6 months later. What are the 3 most likely causes?" |
+| **First Principles** | Architecture, major refactors | "Strip all assumptions. What's the simplest possible solution from ground truth?" |
+| **Inversion** | Risk assessment, refactoring | "How would you guarantee this fails? Now avoid those things." |
+| **Red Team** | Security, APIs, public surfaces | "You're an attacker. How do you break this?" |
+| **Constraint Removal** | Innovation, scope decisions | "Remove all constraints (time, tech, team). What changes? What stays the same?" |
+| **Stakeholder Mapping** | Multi-user features | "Re-evaluate from each stakeholder's perspective. Who loses?" |
+
+### Auto-Selection Rules
+- For spec/plan tasks → Pre-mortem (default)
+- For architecture tasks → First Principles
+- For refactoring tasks → Inversion
+- For security-sensitive → Red Team
+- For scope decisions → Constraint Removal
+
+### Execution
+1. Auto-select the most relevant method based on task type
+2. Apply the method's prompt to the current brainstorm output
+3. Append insights to requirements doc under "## Deepening Insights" section
+4. If insights reveal significant gaps, re-run self-interview for those specific areas
+
 ## Phase 2: Approach Generation
 
 Generate 2-3 concrete approaches based on the answers from Phase 1 and your codebase analysis.

codex/skills/flow-code-brainstorm/SKILL.md

@@ -113,13 +113,12 @@ Proceed to Step 4 immediately.
 
 ## Workflow
 
-Read [steps.md](steps.md) and follow each step in order.
-
-**CRITICAL — Step 4 (Research)**: You MUST launch ALL scouts listed in steps.md in ONE parallel Task call. Do NOT skip scouts or run them sequentially. Each scout provides unique signal.
-
-If review was decided (rp/codex/export):
-- rp or codex: run `/flow-code:plan-review` after Step 9, fix issues until it passes
-- export: run `/flow-code:plan-review` with export mode after Step 9
+Execute steps from `steps/` directory one at a time (JIT loading — only read the current step):
+1. Read `steps/step-01-init.md` and execute
+2. Read `steps/step-02-research.md` and execute
+3. Read `steps/step-03-gap-analysis.md` and execute
+4. Read `steps/step-04-task-breakdown.md` and execute
+5. Read `steps/step-05-output.md` and execute
 
 ## Output