@c-d-cc/reap 0.10.2 → 0.11.0

package/dist/templates/artifacts/01-objective.md

@@ -1,3 +1,4 @@
+# REAP MANAGED — Do not modify directly. Use reap run commands.
 # Objective
 
 ## Goal

package/dist/templates/artifacts/02-planning.md

@@ -1,3 +1,4 @@
+# REAP MANAGED — Do not modify directly. Use reap run commands.
 # Planning
 
 ## Summary

package/dist/templates/artifacts/03-implementation.md

@@ -1,3 +1,4 @@
+# REAP MANAGED — Do not modify directly. Use reap run commands.
 # Implementation Log
 
 ## Completed Tasks

package/dist/templates/artifacts/04-validation.md

@@ -1,3 +1,4 @@
+# REAP MANAGED — Do not modify directly. Use reap run commands.
 # Validation Report
 
 ## Result: pending

package/dist/templates/artifacts/05-completion.md

@@ -1,3 +1,4 @@
+# REAP MANAGED — Do not modify directly. Use reap run commands.
 # Completion
 
 ## Summary

package/dist/templates/artifacts/merge/01-detect.md

@@ -1,3 +1,4 @@
+# REAP MANAGED — Do not modify directly. Use reap run commands.
 # Detect
 
 ## Parents

package/dist/templates/artifacts/merge/02-mate.md

@@ -1,3 +1,4 @@
+# REAP MANAGED — Do not modify directly. Use reap run commands.
 # Mate (Genome Mating)
 
 ## Conflicts

package/dist/templates/artifacts/merge/03-merge.md

@@ -1,3 +1,4 @@
+# REAP MANAGED — Do not modify directly. Use reap run commands.
 # Merge (Source Merge)
 
 ## Git Merge Result

package/dist/templates/artifacts/merge/04-sync.md

@@ -1,3 +1,4 @@
+# REAP MANAGED — Do not modify directly. Use reap run commands.
 # Sync (Genome-Source Consistency)
 
 ## Genome-Source Comparison

package/dist/templates/artifacts/merge/05-validation.md

@@ -1,3 +1,4 @@
+# REAP MANAGED — Do not modify directly. Use reap run commands.
 # Validation
 
 ## Result: pending

package/dist/templates/artifacts/merge/06-completion.md

@@ -1,3 +1,4 @@
+# REAP MANAGED — Do not modify directly. Use reap run commands.
 # Completion
 
 ## Summary

package/dist/templates/commands/reap.abort.md

@@ -2,72 +2,4 @@
 description: "REAP Abort — Abort the current generation and return to initial state"
 ---
 
-# Abort
-
-Abort the current generation and return to initial state.
-
-## Gate (Preconditions)
-- Read `.reap/life/current.yml`
-- If no active Generation: ERROR — "No active Generation to abort." **STOP**
-
-## Steps
-
-### 1. Show Current State
-- Display: generation id, goal, current stage
-- Ask: "이 generation을 abort 하시겠습니까?"
-- If no: **STOP**
-
-### 2. Abort Reason
-- Ask: "abort 사유를 입력해주세요"
-
-### 3. Source Code Handling
-- Check `git diff --name-only` for uncommitted changes
-- If no changes: skip to Step 4
-- If changes exist, present the changed files and ask:
-  - **rollback**: `git checkout .` — 모든 변경 revert
-  - **stash**: `git stash push -m "reap-abort: {gen-id}"` — stash에 저장
-  - **hold**: 변경 유지 (working tree에 그대로)
-
-### 4. Backlog Save
-- Ask: "Goal과 진행 상황을 backlog에 저장할까요? (yes/no)"
-- If yes:
-  - Read `01-objective.md` for goal/spec content
-  - Read `03-implementation.md` if exists for progress
-  - Create `.reap/life/backlog/aborted-{gen-id}.md`:
-    ```markdown
-    ---
-    type: task
-    status: pending
-    aborted: true
-    abortedFrom: {gen-id}
-    abortReason: "{reason}"
-    stage: {current stage}
-    sourceAction: rollback|stash|hold
-    stashRef: "reap-abort: {gen-id}"  # only if stash
-    changedFiles:
-      - {file1}
-      - {file2}
-    ---
-
-    # [Aborted] {goal}
-
-    ## Original Goal
-    {goal from objective}
-
-    ## Progress
-    {stage} 단계에서 중단.
-    {implementation summary if available}
-
-    ## Resume Guide
-    {if stash: "git stash pop으로 코드 복구"}
-    {if hold: "코드 변경이 working tree에 유지됨"}
-    {if rollback: "코드 변경이 revert됨. objective부터 재시작 필요"}
-    ```
-
-### 5. Cleanup
-- Delete all artifact files from `.reap/life/` (`01-*.md` through `05-*.md`)
-- Clear `current.yml` (write empty content)
-- Do NOT record in lineage (incomplete generation)
-
-## Completion
-- "Generation {gen-id} aborted. {backlog saved / not saved}."
+Run `reap run abort` and follow the stdout instructions exactly.

package/dist/templates/commands/reap.back.md

@@ -2,61 +2,4 @@
 description: "REAP Back — Return to a previous lifecycle stage"
 ---
 
-# Return to Previous Stage
-
-## Gate (Preconditions)
-- Read `.reap/life/current.yml`
-- If no active Generation: ERROR — "No active Generation." **STOP**
-- If current stage is `objective` (first stage): ERROR — "Already at the first stage. Cannot go back." **STOP**
-
-## Steps
-
-1. Determine target stage:
-   - `/reap.back` alone: go to the immediately previous stage
-   - `/reap.back [stage]`: go directly to the specified stage
-   - Stage order: objective → planning → implementation → validation → completion
-   - Only backward transitions are allowed
-2. Ask the human for the reason for regression
-3. Collect regression reason and reference info:
-   - **reason**: why we are going back
-   - **refs**: related files/locations (artifact sections, source code locations)
-4. Modify `current.yml`:
-   - Change `stage` to the target stage
-   - Add a regression entry to `timeline`:
-     ```yaml
-     - stage: [target stage]
-       at: [current ISO 8601]
-       from: [original stage]
-       reason: [regression reason]
-       refs:
-         - [ref 1]
-         - [ref 2]
-     ```
-5. Artifact handling:
-   - **Before the target stage**: preserve
-   - **Target stage**: overwrite on re-entry (implementation only: append)
-   - **After the target stage**: preserve, overwrite on re-entry
-6. Record the regression reason at the top of the target stage artifact as a `## Regression` section:
-   ```markdown
-   ## Regression
-   - **From**: [original stage]
-   - **Reason**: [regression reason]
-   - **Refs**: [reference list]
-   - **Affected**: [affected subsequent artifacts]
-   ```
-
-### Hook Execution (Regression)
-7. Scan `.reap/hooks/` for files matching `onLifeRegretted.*`
-   - For each matched file (sorted by `order` from frontmatter, then alphabetically):
-     1. Read the frontmatter (`condition`, `order`)
-     2. Evaluate `condition` by running `.reap/hooks/conditions/{condition}.sh` (exit 0 = met, non-zero = skip):
-        - If `condition` is absent: treat as `always`
-        - If the condition script doesn't exist: warn and skip the hook
-        - Default conditions: `always`, `has-code-changes`, `version-bumped`
-        - Users can add custom conditions by placing scripts in `.reap/hooks/conditions/`
-     3. Execute based on file extension:
-        - `.md`: read the file content (after frontmatter) as AI prompt and follow the instructions
-        - `.sh`: run as shell script in the project root directory
-
-## Completion
-- "Returned to [stage] stage. Proceed with `/reap.[stage]`."
+Run `reap run back` and follow the stdout instructions exactly.

package/dist/templates/commands/reap.completion.md

@@ -1,148 +1,5 @@
 ---
-description: "REAP Completion — Retrospect, evolve the Genome, and finalize the Generation"
+description: "REAP Completion — Retrospect, evolve the Genome, archive, and finalize the Generation"
 ---
 
-# Completion
-
-<HARD-GATE>
-Do NOT apply Genome changes without human confirmation — UNLESS called from `/reap.evolve` (Autonomous Override active).
-Do NOT finalize Genome changes without running Validation Commands.
-</HARD-GATE>
-
-## Gate (Preconditions)
-- Read `.reap/life/current.yml` and verify that stage is `completion`
-- Verify that `.reap/life/04-validation.md` exists
-- If not met: ERROR — state the reason and **STOP**
-
-## Context (Generation Info)
-- Read `.reap/life/current.yml` for the current generation info (id, goal, genomeVersion)
-
-## Steps
-
-### Phase 0: Summary
-
-1. Fill the `## Summary` section of `05-completion.md` using data from `current.yml` and `04-validation.md`:
-   - **Goal**: from `current.yml` goal
-   - **Period**: from `current.yml` startedAt ~ completedAt
-   - **Genome Version**: from `current.yml` genomeVersion (v[N] → v[N+1])
-   - **Result**: from `04-validation.md` result (pass/partial/fail)
-   - **Key Changes**: brief one-line summary of what was delivered
-
-### Phase 1: Retrospective
-
-2. Read all `type: genome-change` and `type: environment-change` items from `.reap/life/backlog/`
-3. Read the deferred task list from `.reap/life/03-implementation.md`
-4. Compile lessons learned from this generation
-   - **Limit**: Maximum 5 lessons. Keep only the most impactful ones.
-5. Compile genome changes to apply (which genome file to modify and how)
-
-### Phase 2: Garbage Collection (Codebase Health)
-
-6. Referencing the Enforced Rules in `.reap/genome/conventions.md`, check codebase consistency:
-   - Check whether new convention violations have been introduced
-   - Identify technical debt incurred during implementation
-7. For each piece of technical debt found, add it to `.reap/life/backlog/`:
-   ```markdown
-   ---
-   type: task
-   status: pending
-   ---
-   # [Title]
-   [Description]
-   ```
-
-### Phase 3: Backlog Cleanup
-
-8. Add deferred tasks to `.reap/life/backlog/` as `type: task`
-9. Also add other next-generation goal candidates to the backlog
-10. Finalize the retrospective with the human
-
-### Phase 4: Genome Application
-
-11. Read `type: genome-change` and `type: environment-change` items from `.reap/life/backlog/`
-12. Apply genome-change items to the corresponding files in `.reap/genome/`:
-    - **Map principle**: Each genome file should be **~100 lines or less**
-    - If exceeding 100 lines, extract into files under `domain/`
-    - **When writing domain/ files, follow the guide in `~/.reap/templates/domain-guide.md`**
-13. Apply environment-change items to the corresponding files in `.reap/environment/`
-14. **Verify**: Run the Validation Commands from constraints.md to confirm genome changes do not conflict with existing code
-15. **Human confirmation**:
-    - **If called from `/reap.evolve`** (Autonomous Override active): Apply genome changes automatically after Validation Commands pass. Do NOT pause for human confirmation.
-    - **If called standalone**: Show the modified genome/environment content to the human and get approval. Do NOT finalize changes until the human approves.
-16. For each applied `type: genome-change` and `type: environment-change` backlog item, update its frontmatter to `status: consumed` and add `consumedBy: gen-XXX-{hash}`
-
-### Phase 5: Hook Suggestion
-
-17. Read the last 3 completed generations from `.reap/lineage/` (sorted by gen number, most recent first)
-    - For each: read `03-implementation.md` (Implementation Notes) and `05-completion.md` (Retrospective)
-    - Identify **manual tasks that were repeated across 2+ generations**
-    - Examples: docs update, lint fix, dependency sync, test data setup, specific file regeneration
-18. If a repeated pattern is found, engage the human with a step-by-step confirmation:
-    a. **Describe the pattern**: "최근 N개 generation에서 '[작업 설명]'이 반복적으로 수행되었습니다."
-    b. **Ask if it should be a hook**: "이 작업을 hook으로 자동화할까요? (yes/no)"
-    c. If yes, ask **event**: "어떤 이벤트에서 실행할까요?"
-       - `onLifeCompleted` — generation 완료 후
-       - `onLifeTransited` — stage 전환 시
-       - `onLifeStarted` — generation 시작 시
-       - `onLifeRegretted` — stage 회귀 시
-    d. Ask **condition**: "실행 조건은 무엇인가요?"
-       - `always` — 항상
-       - `has-code-changes` — src/ 변경이 있을 때
-       - `version-bumped` — version bump가 있을 때
-       - Custom — 유저가 직접 기술
-    e. Ask **hook name**: "hook 이름을 지어주세요 (예: lint-fix, docs-sync)"
-    f. **Preview**: 생성될 hook 파일 내용을 보여주고 확인:
-       ```
-       파일: .reap/hooks/{event}.{name}.md
-       ---
-       condition: {condition}
-       order: 50
-       ---
-       {작업 내용}
-       ```
-    g. 유저 확인 후 `.reap/hooks/{event}.{name}.md` 생성
-19. 반복 패턴이 없으면 skip — "반복 패턴이 감지되지 않았습니다."
-20. **Limit**: 한 번에 최대 2개까지만 제안 (과부하 방지)
-
-### Phase 6: Lineage Compression
-
-21. Check if lineage compression is needed:
-    - Count total lines in `.reap/lineage/` and number of generations
-    - **Level 1 trigger**: total lines > 5,000 AND generations >= 5
-    - **Level 2 trigger**: Level 1 compressed `.md` files > 100
-22. If Level 1 triggered:
-    - Compress oldest uncompressed generation directories into single `.md` files
-    - Protect: recent 3 generations + DAG leaf nodes
-    - Preserve DAG metadata in frontmatter (id, parents, genomeHash)
-23. If Level 2 triggered:
-    - Run `git fetch --all` to update remote refs
-    - Scan all branches (local + remote) for fork points
-    - Compress eligible Level 1 files into single `epoch.md` (append if exists)
-    - Protect: recent 9 Level 1 files + all generations at/after fork points
-    - epoch.md frontmatter contains `generations` array with hash chain (id, parents, genomeHash)
-24. Report compression results: "Compressed N generations (Level 1: X, Level 2: Y)"
-    - If no compression needed: skip silently
-
-## Self-Verification
-Before saving the artifact, verify:
-- [ ] Are lessons concrete and applicable to the next generation? (No vague "do better next time")
-- [ ] Have genome changes been confirmed by the human?
-- [ ] Do Validation Commands still pass after the changes?
-- [ ] Have deferred tasks been added to the backlog?
-
-❌ Bad lesson: "We should write more tests"
-✅ Good lesson: "SSE streaming responses are difficult to unit test, so prioritize integration tests"
-
-## Artifact Generation (Progressive Recording)
-- **Language**: Write all artifact content in the user's configured language (see REAP Guide § Language).
-- **Immediately upon entering this stage**: Read `~/.reap/templates/05-completion.md` and create `.reap/life/05-completion.md` with empty sections ready to fill
-- **Update incrementally as each phase completes**:
-  - After Phase 1 (Retrospective) → fill in Lessons Learned, Genome Change Proposals
-  - After Phase 2 (Garbage Collection) → update with any tech debt findings
-  - After Phase 3 (Backlog Cleanup) → fill in Deferred Task Handoff, Next Generation Backlog
-  - After Phase 4 (Genome Application) → fill in Genome Changelog sections
-- The artifact should reflect the current state of completion work at all times
-
-## Completion
-- **If called from `/reap.evolve`** (Autonomous Override active): Proceed automatically to `/reap.next`.
-- **If called standalone**: "Finalize the generation with `/reap.next`. Lineage archiving will be performed automatically."
+Run `reap run completion` and follow the stdout instructions exactly.

package/dist/templates/commands/reap.evolve.md

@@ -2,54 +2,4 @@
 description: "REAP Evolve — Run the full lifecycle for a Generation"
 ---
 
-# Evolve (Full Lifecycle)
-
-Run the entire Generation lifecycle from start to finish, interactively with the human.
-
-<HARD-GATE>
-NEVER modify `current.yml` directly to change the stage. ALWAYS use `/reap.next` to advance and `/reap.back` to regress.
-Direct modification skips artifact creation and breaks the lifecycle. This is non-negotiable.
-
-Specifically PROHIBITED:
-- Running `node -e` or any script that writes to `current.yml`
-- Using YAML.stringify() to update the stage field
-- Any method other than invoking `/reap.next` or `/reap.back` slash commands
-
-The ONLY legitimate way to change the stage is through `/reap.next` (forward) or `/reap.back` (regression).
-</HARD-GATE>
-
-## Gate (Preconditions)
-- Read `.reap/life/current.yml`
-- If no active Generation exists: run `/reap.start` first to create one
-- If an active Generation exists: resume from the current stage
-
-## Autonomous Override
-When `/reap.evolve` calls each stage command, the following overrides apply:
-- **Skip routine human confirmations**: Do NOT pause to show artifacts and ask for approval at the end of each stage. Proceed autonomously.
-- **Skip environment/genome interactive setup questions**: Use existing data, fill in what you can, skip what's empty.
-- **STOP only when genuinely blocked**: ambiguous goal with multiple valid interpretations, uncertain technical decision with significant trade-offs, conflicts in the genome, or unexpected errors.
-- **Escalation sections still apply**: If a stage's Escalation rules trigger, STOP and ask.
-- This override does NOT apply when stages are invoked standalone (e.g., user runs `/reap.objective` directly).
-
-## Lifecycle Loop
-
-Execute the following loop until the generation is complete:
-
-1. Read `current.yml` to determine the current stage
-2. Execute the corresponding stage command:
-   - `objective` → `/reap.objective`
-   - `planning` → `/reap.planning`
-   - `implementation` → `/reap.implementation`
-   - `validation` → `/reap.validation`
-   - `completion` → `/reap.completion`
-3. When the stage command completes, run `/reap.next` to advance
-4. If `/reap.next` archives the generation (from completion), the loop ends
-5. Otherwise, return to step 1
-
-## Handling Issues
-- If validation fails: `/reap.back` to return to implementation (or earlier), then resume the loop
-- If the human wants to pause: stop the loop, the generation state is preserved in `current.yml`
-- If the human wants to skip a stage: advance with `/reap.next` without running the stage command
-
-## Completion
-- "Generation [id] completed. All artifacts archived to lineage."
+Run `reap run evolve` and follow the stdout instructions exactly.

package/dist/templates/commands/reap.help.md

@@ -2,93 +2,4 @@
 description: "REAP Help — Contextual help based on current state"
 ---
 
-# Help
-
-**ARGUMENTS**: $ARGUMENTS
-
-## File Access Rules
-- ONLY read `.reap/life/current.yml` and `.reap/config.yml`. No other files for basic help.
-- For topic help: also read `reap.{name}.md` from the same directory as this file.
-- Do NOT access global node_modules. Do NOT guess file paths.
-
-## If ARGUMENTS is provided → skip to "Topic Help" section below.
-
-## Basic Help (no arguments)
-
-Read `.reap/life/current.yml` and `.reap/config.yml`, then output EXACTLY this format with values filled in. Use markdown formatting. Do NOT add extra explanation.
-
----
-
-**REAP** — Recursive Evolutionary Autonomous Pipeline
-
-AI와 사람이 협업하여 소프트웨어를 세대(Generation) 단위로 진화시키는 개발 파이프라인입니다.
-각 세대는 5단계 lifecycle을 거칩니다:
-
-`Objective` → `Planning` → `Implementation` → `Validation` → `Completion`
-
-Genome(`.reap/genome/`)에 프로젝트의 설계 원칙과 규칙을 기록하고, 세대를 거듭하며 점진적으로 진화합니다.
-
----
-
-[If active generation exists, show:]
-> 🔄 **{id}** — {goal}
-> Stage: `{stage}` → 다음: `/reap.{stage}` 또는 `/reap.next`
-
-[If no active generation, show:]
-> No active Generation → `/reap.start` 또는 `/reap.evolve`
-
----
-
-### 커맨드
-
-| Command | Description |
-|---------|-------------|
-| `/reap.start` | 새 Generation을 시작하고 goal을 설정 |
-| `/reap.evolve` | ⚡ 전체 lifecycle을 자율적으로 실행 (사람 개입 최소화) |
-| `/reap.objective` | 이번 Generation의 목표, 요구사항, 완료 기준을 정의 |
-| `/reap.planning` | 태스크를 분해하고 구현 계획을 수립 |
-| `/reap.implementation` | 계획에 따라 코드를 구현하고 진행 상황을 기록 |
-| `/reap.validation` | 테스트 실행, 완료 기준 충족 여부를 검증 |
-| `/reap.completion` | 회고, Genome 변경 반영, Generation 마무리 |
-| `/reap.next` | 다음 stage로 전진 (completion → archiving + commit) |
-| `/reap.back` | 이전 stage로 회귀 (regression 사유 기록) |
-| `/reap.status` | 프로젝트 상태, Generation 진행도, backlog, Genome 건강도 표시 |
-| `/reap.sync` | 현재 소스코드와 Genome 간 차이를 분석하고 동기화 |
-| `/reap.update` | REAP 최신 버전 확인 및 업그레이드 |
-| `/reap.help` | 도움말. `/reap.help {topic}`으로 주제별 상세 설명 |
-| | **Collaboration** |
-| `/reap.pull {branch}` | fetch + divergence 감지 + merge generation 자동 실행 |
-| `/reap.push` | REAP 상태 검증 + git push |
-| `/reap.merge {branch}` | merge generation 전체 lifecycle 실행 |
-| `/reap.merge.start` | merge generation 생성 + detect |
-| `/reap.merge.evolve` | merge 6단계 자율 실행 |
-
-💡 `/reap.help {topic}` — workflow, genome, backlog, strict, agents, hooks, config, evolve, regression, merge, pull, push, author ...
-
-**Config**: Strict: {on/off} · Auto-Update: {on/off} · Language: {value}
-
----
-
-## Topic Help
-
-If ARGUMENTS contains a topic, look up from the list below. **If NOT in the list, respond: "Unknown topic: '{topic}'. `/reap.help`로 available topics를 확인하세요."**
-
-For command-name topics: read `reap.{name}.md` from the same directory as this file, then explain.
-
-#### Topics
-
-- **workflow** / **lifecycle** — 5-stage lifecycle: objective → planning → implementation → validation → completion. Gate → Steps → Artifact. `/reap.next`로 전진, `/reap.back`으로 회귀.
-- **genome** — `.reap/genome/`: principles.md, conventions.md, constraints.md, domain/*.md. 불변 원칙: 현재 세대에서 직접 수정 금지 → backlog → completion에서 반영.
-- **backlog** — `.reap/life/backlog/`. genome-change, environment-change, task. status: pending/consumed. consumed → lineage, pending → 이월.
-- **strict** — `strict: true`: no generation = 코드 수정 차단, non-implementation = 차단, implementation = plan scope만. Escape: "override"/"bypass strict".
-- **agents** — AgentAdapter 추상화. auto-detect (Claude Code, OpenCode). config.yml `agents` 오버라이드.
-- **hooks** — config.yml: onLifeStarted, onLifeTransited, onLifeCompleted, onLifeRegretted. command/prompt 타입.
-- **config** — `.reap/config.yml`: version, project, entryMode, strict, language, agents, autoUpdate, hooks.
-- **evolve** — `/reap.evolve`: 전체 lifecycle 자율 실행. 사람 확인 없이 objective~completion 순차 진행.
-- **regression** — `/reap.back`: 이전 stage 회귀. timeline + artifact에 Regression 섹션 기록.
-- **minor-fix** — 5분 이내, 설계 변경 없는 수정. stage 전환 없이 현재 artifact에 기록.
-- **compression** — 10,000줄 + 5세대 이상 시 lineage 자동 압축. L1(40줄), L2(60줄).
-- **merge** / **collaboration** — 분산 협업: genome-first merge 6단계 (detect→mate→merge→sync→validation→completion). `/reap.pull`로 fetch+merge, `/reap.push`로 검증+push. 상세: `domain/collaboration.md`, `domain/merge-lifecycle.md`.
-- **author** — HyeonIL Choi. Email: hichoi@c-d.cc / Homepage: https://c-d.cc / LinkedIn: https://www.linkedin.com/in/hichoi-dev / GitHub: https://github.com/casamia918
-- **start** / **objective** / **planning** / **implementation** / **validation** / **completion** / **next** / **back** / **sync** / **status** / **update** / **help** — Read `reap.{name}.md` from same directory, explain.
-- **pull** / **push** / **merge.start** / **merge.detect** / **merge.mate** / **merge.merge** / **merge.sync** / **merge.validation** / **merge.completion** / **merge.evolve** — Read `reap.{name}.md` from same directory, explain.
+Run `reap run help` and follow the stdout instructions exactly.

package/dist/templates/commands/reap.implementation.md

@@ -2,101 +2,4 @@
 description: "REAP Implementation — Implement code through AI+Human collaboration"
 ---
 
-# Implementation
-
-<HARD-GATE>
-Do NOT write code without the task list from 02-planning.md.
-Do NOT write code that violates the Genome (conventions.md, constraints.md).
-Do NOT modify the Genome directly — if you discover an issue, record it in the backlog.
-Do NOT modify the Environment directly — if you discover a change, record it in the backlog.
-</HARD-GATE>
-
-## Gate (Preconditions)
-- Read `.reap/life/current.yml` and verify that stage is `implementation`
-- Verify that `.reap/life/02-planning.md` exists
-- Check git working tree status (`git status`)
-  - If there are uncommitted changes that overlap with files this generation will modify:
-    - Ask the user: "There are uncommitted changes. Would you like to commit them before proceeding? (yes/no/show diff)"
-    - If yes: help the user commit the changes, then proceed
-    - If no: proceed with caution, noting the existing changes
-    - If show diff: show the diff, then ask again
-  - If there are uncommitted changes in unrelated files: proceed normally
-  - If the working tree is clean: proceed normally
-- If not met: ERROR — state the reason and **STOP**
-
-## Context (Generation Info)
-- Read `.reap/life/current.yml` for the current generation info (id, goal, genomeVersion)
-- Read `.reap/genome/conventions.md` — rules to reference continuously during implementation
-- Read `.reap/genome/constraints.md` — technical constraints
-
-## Re-entry Check
-- If `.reap/life/03-implementation.md` already exists, this is a **re-entry due to regression**
-- Read the existing log; if a `## Regression` section exists, understand the regression reason
-- On regression re-entry, **append** to the existing log (preserve existing completion records)
-
-## Steps
-
-### 1. Load Task List
-- Read the task list from `.reap/life/02-planning.md`
-- If `03-implementation.md` already exists, read it (preserve existing records)
-- Identify incomplete (`[ ]`) tasks
-
-### 2. Sequential Implementation
-- Implement starting from the first incomplete task, in order
-- **After EACH task completion**, immediately add it to the Completed Tasks table in `03-implementation.md`
-- **After EACH deferred task decision**, immediately add it to the Deferred Tasks table
-- **After EACH genome issue discovery**, immediately add it to the Genome-Change Backlog Items table
-- Do NOT batch-write the log at the end. Update it as you go.
-- **Strictly** follow the rules in conventions.md
-- **Strictly** follow the technical constraints in constraints.md
-
-### 3. When Genome/Environment Changes Are Discovered
-- If you discover something that must be implemented differently from the spec:
-  a. Record it in `.reap/life/backlog/`:
-     ```markdown
-     ---
-     type: genome-change
-     status: pending
-     target: genome/[relevant file]
-     ---
-     # [Title]
-     [What is different and how it should be changed]
-     ```
-  b. Mark tasks that depend on this change as `[deferred]` in `02-planning.md`
-  c. Record the reason for deferral
-- **Do NOT modify Genome or Environment files directly. This is non-negotiable.**
-
-### 3b. When Out-of-Scope Issues Are Discovered
-- If you discover issues outside the current generation's scope (outdated tests, tech debt, broken code unrelated to current tasks, etc.):
-  a. Record it in `.reap/life/backlog/`:
-     ```markdown
-     ---
-     type: task
-     status: pending
-     ---
-     # [Title]
-     [Description of the issue and what needs to be done]
-     ```
-  b. Do NOT fix it in the current generation unless the human explicitly approves
-  c. Record it in `03-implementation.md` under "Genome-Change Backlog Items" or "Implementation Notes"
-
-### 4. Completion Marking
-- Mark completed tasks as `[x]` in `02-planning.md`
-
-## Escalation
-In the following situations, **STOP and ask the human**:
-- When task requirements are unclear
-- When an architectural decision is needed but not covered in the plan
-- When a conflict arises with existing code
-- When the scope turns out to be significantly larger than expected
-
-## Artifact Generation (Progressive Recording)
-- **Language**: Write all artifact content in the user's configured language (see REAP Guide § Language).
-- **Immediately upon entering this stage**: Read `~/.reap/templates/03-implementation.md` and create `.reap/life/03-implementation.md` with empty tables ready to fill (or preserve existing content on re-entry)
-- **Update continuously during Step 2**: After each task, deferral, or discovery, update the artifact immediately
-- By the time all tasks are done, the artifact should already be complete — no separate "generation" step is needed
-- Add Implementation Notes at the end summarizing notable decisions
-
-## Completion
-- "Proceed to the Validation stage with `/reap.next`."
-- "If issues are found during Validation, you can return with `/reap.back`."
+Run `reap run implementation` and follow the stdout instructions exactly.

package/dist/templates/commands/reap.merge.completion.md

@@ -1,20 +1,5 @@
 ---
-description: "REAP Merge Completion — Finalize the merge generation"
+description: "REAP Merge Completion — Finalize the merge generation with archiving"
 ---
 
-# Merge Completion
-
-Finalize the merge generation after validation passes.
-
-## Gate
-- Verify current generation is type: merge and stage: completion
-- Verify `05-validation.md` exists
-
-## Steps
-
-1. Commit the merged source + genome changes
-2. Write `06-completion.md` with:
-   - Summary of what was merged
-   - Genome changes applied
-   - Lessons learned
-3. Proceed with `/reap.next` to archive
+Run `reap run merge-completion` and follow the stdout instructions exactly.

package/dist/templates/commands/reap.merge.detect.md

@@ -2,19 +2,4 @@
 description: "REAP Merge Detect — Analyze divergence between branches"
 ---
 
-# Merge Detect
-
-Analyze the divergence between the current branch and the target branch.
-
-## Gate
-- Verify current generation is type: merge and stage: detect
-
-## Steps
-
-1. Read `01-detect.md` (created by `reap merge` CLI)
-2. Review the divergence report with the human:
-   - Common ancestor
-   - Genome changes on each side
-   - Conflicts (WRITE-WRITE, CROSS-FILE)
-3. If the detect needs to be re-run, use `reap merge {branch}` again
-4. When satisfied, proceed with `/reap.next`
+Run `reap run merge-detect` and follow the stdout instructions exactly.