selfish-pipeline 1.2.1 → 1.3.0

package/.claude-plugin/marketplace.json

@@ -6,14 +6,14 @@
   },
   "metadata": {
     "description": "Automated pipeline for Claude Code — spec → plan → tasks → implement → review → clean",
-    "version": "1.2.1"
+    "version": "1.3.0"
   },
   "plugins": [
     {
       "name": "selfish",
       "source": "./",
       "description": "Automated pipeline for Claude Code. Automates the full development cycle: spec → plan → tasks → implement → review → clean.",
-      "version": "1.2.1",
+      "version": "1.3.0",
       "category": "automation",
       "tags": ["pipeline", "automation", "spec", "plan", "implement", "review", "critic-loop"]
     }

package/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "selfish",
-  "version": "1.2.1",
+  "version": "1.3.0",
   "description": "Automated pipeline for Claude Code. Automates the full development cycle: spec → plan → tasks → implement → review → clean.",
   "author": { "name": "jhlee0409", "email": "relee6203@gmail.com" },
   "homepage": "https://github.com/jhlee0409/selfish-pipeline",

package/MIGRATION.md

@@ -82,12 +82,15 @@ For new projects, you can auto-generate it with `/selfish:init`.
 ## What Stays the Same
 
 - `.claude/selfish.config.md` file format and path
-- `specs/{feature}/` artifact paths
-- `memory/` references (checkpoint, principles, research, decisions)
 - `.selfish-*` state file paths
 - `git tag selfish/pre-*` safety tags
 - Internal logic of hook scripts
 
+## What Changed (v1.2.2+)
+
+- `specs/{feature}/` → `.claude/selfish/specs/{feature}/`
+- `memory/` → `.claude/selfish/memory/` (checkpoint, principles, research, decisions)
+
 ## FAQ
 
 **Q: Do I need to recreate `.claude/selfish.config.md` for existing projects?**

package/agents/selfish-architect.md

@@ -3,6 +3,7 @@ name: selfish-architect
 description: "Architecture analysis agent — remembers ADR decisions and architecture patterns across sessions to provide consistent design guidance."
 tools:
   - Read
+  - Write
   - Grep
   - Glob
   - Bash

package/commands/analyze.md

@@ -34,7 +34,7 @@ config.architecture = the architecture pattern used in the project
 
 ### 1. Load Artifacts
 
-From `specs/{feature}/`:
+From `.claude/selfish/specs/{feature}/`:
 - **spec.md** (required)
 - **plan.md** (required)
 - **tasks.md** (if present)
@@ -66,7 +66,7 @@ Validate across 6 categories:
 - Mismatches between technical decisions in plan and execution in tasks
 
 #### E. Principles Compliance (PRINCIPLES)
-- Validate against MUST principles in memory/principles.md if present
+- Validate against MUST principles in .claude/selfish/memory/principles.md if present
 - Potential violations of {config.architecture} rules
 
 #### F. Unidentified Risks (RISK)

package/commands/architect.md

@@ -1,12 +1,13 @@
 ---
 name: selfish:architect
-description: "Architecture analysis and design advice (read-only)"
+description: "Architecture analysis and design advice"
 argument-hint: "[analysis target or design question]"
 disable-model-invocation: true
 context: fork
 agent: selfish-architect
 allowed-tools:
   - Read
+  - Write
   - Grep
   - Glob
   - Bash
@@ -18,7 +19,7 @@ model: sonnet
 # /selfish:architect — Architecture Analysis and Design Advice
 
 > Analyzes the codebase architecture and records design decisions.
-> Ensures design quality through convergence-based Critic Loop. **Read-only** — does not modify code.
+> Ensures design quality through convergence-based Critic Loop. Does not modify source code. May write ADR files to `.claude/selfish/memory/decisions/`.
 
 ## Arguments
 
@@ -91,7 +92,7 @@ Structure analysis results and **print to console**:
 
 ### 4. Critic Loop
 
-> **Always** read `docs/critic-loop-rules.md` first and follow it.
+> **Always** read `${CLAUDE_PLUGIN_ROOT}/docs/critic-loop-rules.md` first and follow it.
 
 Run the critic loop until convergence. Safety cap: 7 passes.
 
@@ -110,7 +111,7 @@ Run the critic loop until convergence. Safety cap: 7 passes.
 
 ### 5. Save ADR (for design decisions)
 
-If ADR type, save to `memory/decisions/{YYYY-MM-DD}-{topic}.md`:
+If ADR type, save to `.claude/selfish/memory/decisions/{YYYY-MM-DD}-{topic}.md`:
 
 ```markdown
 # ADR: {title}
@@ -135,7 +136,7 @@ Architecture analysis complete
 
 ## Notes
 
-- **Read-only**: Does not modify code. Performs analysis and suggestions only.
+- **No source modification**: Does not modify project source code. May write ADR files to `.claude/selfish/memory/decisions/`.
 - **Based on actual code**: Explore the actual codebase, not assumptions.
 - **Architecture first**: All suggestions respect {config.architecture} rules.
 - **Incremental changes**: Prefer incremental improvements over big-bang refactoring.

package/commands/auto.md

@@ -34,7 +34,7 @@ If config file is missing: print "`.claude/selfish.config.md` not found. Create
 
 ## Critic Loop Rules (common to all phases)
 
-> **Always** read `docs/critic-loop-rules.md` first and follow it.
+> **Always** read `${CLAUDE_PLUGIN_ROOT}/docs/critic-loop-rules.md` first and follow it.
 > Core: minimum 1 concern per criterion + mandatory Adversarial failure scenario each pass + quantitative evidence required. "PASS" as a single word is prohibited. Uses convergence-based termination with 4 verdicts (PASS/FAIL/ESCALATE/DEFER). On ESCALATE: pause and present options to user even in auto mode.
 
 ---
@@ -60,7 +60,7 @@ If config file is missing: print "`.claude/selfish.config.md` not found. Create
    - Stop Gate Hook activated (blocks response termination on CI failure)
    - File change tracking started
    - Timeline log: `"${CLAUDE_PLUGIN_ROOT}/scripts/selfish-pipeline-manage.sh" log pipeline-start "Auto pipeline: {feature}"`
-5. Create `specs/{feature}/` directory → **record path as `PIPELINE_ARTIFACT_DIR`** (for Clean scope)
+5. Create `.claude/selfish/specs/{feature}/` directory → **record path as `PIPELINE_ARTIFACT_DIR`** (for Clean scope)
 6. Start notification:
    ```
    Auto pipeline started: {feature}
@@ -75,10 +75,10 @@ If config file is missing: print "`.claude/selfish.config.md` not found. Create
 Execute `/selfish:spec` logic inline:
 
 1. Explore codebase for related code (Glob, Grep) — explore by `{config.architecture}` layer
-2. Create `specs/{feature}/spec.md`
+2. Create `.claude/selfish/specs/{feature}/spec.md`
 3. `[NEEDS CLARIFICATION]` items are **auto-resolved with best-guess** (clarify skipped)
    - Tag auto-resolved items with `[AUTO-RESOLVED]`
-4. **Retrospective check**: if `memory/retrospectives/` exists, load and check:
+4. **Retrospective check**: if `.claude/selfish/memory/retrospectives/` exists, load and check:
    - Were there previous `[AUTO-RESOLVED]` items that turned out wrong? Flag similar patterns.
    - Were there scope-related issues in past specs? Warn about similar ambiguities.
 5. **Critic Loop until convergence** (safety cap: 5, follow Critic Loop rules):
@@ -97,7 +97,7 @@ Execute `/selfish:plan` logic inline:
 
 1. Load spec.md
 2. If technical uncertainties exist → auto-resolve via WebSearch/code exploration → create research.md
-3. Create `specs/{feature}/plan.md`
+3. Create `.claude/selfish/specs/{feature}/plan.md`
    - **If setting numerical targets (line counts etc.), include structure-analysis-based estimates** (e.g., "function A ~50 lines, component B ~80 lines → total ~130 lines")
 4. **Critic Loop until convergence** (safety cap: 7, follow Critic Loop rules):
    - Criteria: COMPLETENESS, FEASIBILITY, ARCHITECTURE, RISK, PRINCIPLES
@@ -124,14 +124,14 @@ Execute `/selfish:tasks` logic inline:
    - Validate dependency graph is a DAG (no circular references)
    - [P] tasks **must be executed in parallel** in Phase 4 (declaring [P] then running sequentially is prohibited)
 4. Coverage mapping (FR → Task)
-5. **Retrospective check**: if `memory/retrospectives/` exists, load and check:
+5. **Retrospective check**: if `.claude/selfish/memory/retrospectives/` exists, load and check:
    - Were there previous parallel conflict issues ([P] file overlaps)? Flag similar file patterns.
    - Were there tasks that were over-decomposed or under-decomposed? Adjust granularity.
 6. **Critic Loop until convergence** (safety cap: 5, follow Critic Loop rules):
    - COVERAGE: is every FR/NFR mapped to at least 1 task?
    - DEPENDENCIES: is the dependency graph a valid DAG? Do [P] tasks have no file overlaps?
    - FAIL → auto-fix and continue. ESCALATE → pause, present options, resume after response. DEFER → record reason, mark clean.
-7. Create `specs/{feature}/tasks.md`
+7. Create `.claude/selfish/specs/{feature}/tasks.md`
 8. Progress: `✓ 3/6 Tasks complete (tasks: {N}, parallel: {N}, Critic: converged ({N} passes, {M} fixes, {E} escalations))`
 
 ### Phase 4: Implement (4/6)
@@ -171,12 +171,12 @@ Execute `/selfish:implement` logic inline with **dependency-aware orchestration*
    // 5. Read all worker results before proceeding to gate
    ```
 
-6. Perform **3-step gate** on each Implementation Phase completion — **always** read `docs/phase-gate-protocol.md` first. Cannot advance to next phase without passing the gate.
+6. Perform **3-step gate** on each Implementation Phase completion — **always** read `${CLAUDE_PLUGIN_ROOT}/docs/phase-gate-protocol.md` first. Cannot advance to next phase without passing the gate.
    - On gate pass: create phase rollback point `"${CLAUDE_PLUGIN_ROOT}/scripts/selfish-pipeline-manage.sh" phase-tag {phase_number}`
 7. Real-time `[x]` updates in tasks.md
 8. After full completion, run `{config.ci}` final verification
    - On pass: `"${CLAUDE_PLUGIN_ROOT}/scripts/selfish-pipeline-manage.sh" ci-pass` (releases Stop Gate)
-9. **Implement retrospective**: if unexpected problems arose that weren't predicted in Plan, record in `specs/{feature}/retrospective.md` (for memory update in Clean)
+9. **Implement retrospective**: if unexpected problems arose that weren't predicted in Plan, record in `.claude/selfish/specs/{feature}/retrospective.md` (for memory update in Clean)
 10. Progress: `✓ 4/6 Implement complete ({completed}/{total} tasks, CI: ✓, Mini-Review: ✓, Checkpoint: ✓)`
 
 ### Phase 5: Review (5/6)
@@ -187,7 +187,7 @@ Execute `/selfish:review` logic inline:
 
 1. Review implemented changed files (`git diff HEAD`)
 2. Check code quality, `{config.architecture}` rules, security, performance, `{config.code_style}` pattern compliance
-3. **Retrospective check**: if `memory/retrospectives/` exists, load and check:
+3. **Retrospective check**: if `.claude/selfish/memory/retrospectives/` exists, load and check:
    - Were there recurring Critical finding categories in past reviews? Prioritize those perspectives.
    - Were there false positives that wasted effort? Reduce sensitivity for those patterns.
 4. **Critic Loop until convergence** (safety cap: 5, follow Critic Loop rules):
@@ -206,8 +206,8 @@ Execute `/selfish:review` logic inline:
 Artifact cleanup and codebase hygiene check after implementation and review:
 
 1. **Artifact cleanup** (scope-limited):
-   - **Delete only the `specs/{feature}/` directory created by the current pipeline**
-   - If other `specs/` subdirectories exist, **do not delete them** (only inform the user of their existence)
+   - **Delete only the `.claude/selfish/specs/{feature}/` directory created by the current pipeline**
+   - If other `.claude/selfish/specs/` subdirectories exist, **do not delete them** (only inform the user of their existence)
    - Do not leave pipeline intermediate artifacts in the codebase
 2. **Dead code scan**:
    - Detect unused imports from the implementation process (check with `{config.ci}`)
@@ -217,12 +217,12 @@ Artifact cleanup and codebase hygiene check after implementation and review:
    - Run `{config.ci}` final execution
    - Auto-fix on failure (max 2 attempts)
 4. **Memory update** (if applicable):
-   - Reusable patterns found during pipeline → record in `memory/`
-   - If there were `[AUTO-RESOLVED]` items → record decisions in `memory/decisions/`
-   - **If retrospective.md exists** → record as patterns missed by the Plan phase Critic Loop in `memory/retrospectives/` (reuse as RISK checklist items in future runs)
-   - **If review-report.md exists** → copy to `memory/reviews/{feature}-{date}.md` before specs/ deletion
+   - Reusable patterns found during pipeline → record in `.claude/selfish/memory/`
+   - If there were `[AUTO-RESOLVED]` items → record decisions in `.claude/selfish/memory/decisions/`
+   - **If retrospective.md exists** → record as patterns missed by the Plan phase Critic Loop in `.claude/selfish/memory/retrospectives/` (reuse as RISK checklist items in future runs)
+   - **If review-report.md exists** → copy to `.claude/selfish/memory/reviews/{feature}-{date}.md` before .claude/selfish/specs/ deletion
 5. **Quality report** (structured pipeline metrics):
-   - Generate `memory/quality-history/{feature}-{date}.json` with the following structure:
+   - Generate `.claude/selfish/memory/quality-history/{feature}-{date}.json` with the following structure:
      ```json
      {
        "feature": "{feature}",
@@ -237,9 +237,9 @@ Artifact cleanup and codebase hygiene check after implementation and review:
        "totals": { "changed_files": N, "auto_resolved": N, "escalations": N }
      }
      ```
-   - Create `memory/quality-history/` directory if it does not exist
+   - Create `.claude/selfish/memory/quality-history/` directory if it does not exist
 6. **Checkpoint reset**:
-   - Clear `memory/checkpoint.md` (pipeline complete = session goal achieved)
+   - Clear `.claude/selfish/memory/checkpoint.md` (pipeline complete = session goal achieved)
 7. **Timeline finalize**:
    ```bash
    "${CLAUDE_PLUGIN_ROOT}/scripts/selfish-pipeline-manage.sh" log pipeline-end "Pipeline complete: {feature}"
@@ -267,7 +267,7 @@ Auto pipeline complete: {feature}
 ├─ Changed files: {N}
 ├─ Auto-resolved: {N} (review recommended)
 ├─ Retrospective: {present/absent}
-└─ specs/{feature}/ cleaned up
+└─ .claude/selfish/specs/{feature}/ cleaned up
 ```
 
 ## Abort Conditions
@@ -284,8 +284,8 @@ Pipeline aborted (Phase {N}/6)
 ├─ Reason: {abort cause}
 ├─ Completed phases: {completed list}
 ├─ Rollback: git reset --hard selfish/pre-auto (restores state before implementation)
-├─ Checkpoint: memory/checkpoint.md (last phase gate passed)
-├─ Artifacts: specs/{feature}/ (partial completion, manual deletion needed if Clean did not run)
+├─ Checkpoint: .claude/selfish/memory/checkpoint.md (last phase gate passed)
+├─ Artifacts: .claude/selfish/specs/{feature}/ (partial completion, manual deletion needed if Clean did not run)
 └─ Resume: /selfish:resume → /selfish:implement (checkpoint-based)
 ```
 

package/commands/checkpoint.md

@@ -13,7 +13,7 @@ allowed-tools:
 
 # /selfish:checkpoint — Save Session State
 
-> Saves the current work state to memory/checkpoint.md.
+> Saves the current work state to .claude/selfish/memory/checkpoint.md.
 > Preserves progress even if the session is interrupted.
 
 ## Arguments
@@ -41,7 +41,7 @@ Collect automatically:
 
 ### 2. Save Checkpoint
 
-**Overwrite** `memory/checkpoint.md` (keep only the latest state):
+**Overwrite** `.claude/selfish/memory/checkpoint.md` (keep only the latest state):
 
 ````markdown
 # Session Checkpoint

package/commands/clarify.md

@@ -22,7 +22,7 @@ model: sonnet
 
 ### 1. Load Spec
 
-1. Read `specs/{feature}/spec.md` — stop if not found
+1. Read `.claude/selfish/specs/{feature}/spec.md` — stop if not found
 2. If a `[NEEDS CLARIFICATION]` section exists, process it first
 3. Quickly check existing codebase for related patterns
 

package/commands/debug.md

@@ -61,7 +61,7 @@ Verify starting from highest probability.
 
 ### 5. Critic Loop
 
-> **Always** read `docs/critic-loop-rules.md` first and follow it.
+> **Always** read `${CLAUDE_PLUGIN_ROOT}/docs/critic-loop-rules.md` first and follow it.
 
 Run the critic loop until convergence. Safety cap: 5 passes.
 
@@ -84,7 +84,22 @@ Run the critic loop until convergence. Safety cap: 5 passes.
 
 Retry after fixing on failure (max 3 attempts).
 
-### 7. Final Output
+### 7. Retrospective Entry (if new pattern found)
+
+If this debug session reveals a pattern not previously documented in `.claude/selfish/memory/retrospectives/`:
+
+Append to `.claude/selfish/memory/retrospectives/{YYYY-MM-DD}.md`:
+```markdown
+## Pattern: {category}
+**What happened**: {concrete description}
+**Root cause**: {why this bug occurred}
+**Prevention rule**: {actionable rule — usable in future plan/implement phases}
+**Severity**: Critical | Warning
+```
+
+Only write if the pattern is new and actionable. Generic observations are prohibited.
+
+### 8. Final Output
 
 ```
 Debug complete

package/commands/doctor.md

@@ -60,7 +60,7 @@ Run ALL checks regardless of earlier failures. Do not short-circuit.
 |-------|-----|------|------|
 | Global CLAUDE.md exists | Read `~/.claude/CLAUDE.md` | File exists | ⚠ Warning: no global CLAUDE.md. Selfish skills won't auto-trigger from intent. Fix: run `/selfish:init` |
 | SELFISH block present | Grep for `<!-- SELFISH:START -->` and `<!-- SELFISH:END -->` in `~/.claude/CLAUDE.md` | Both markers found | Fix: run `/selfish:init` to inject SELFISH block |
-| SELFISH block version | Extract version from `<!-- SELFISH:VERSION:X.Y.Z -->` | Version matches or is newer than plugin version | ⚠ Warning: SELFISH block is outdated (found {old}, current {new}). Fix: run `/selfish:init` to update |
+| SELFISH block version | Extract version from `<!-- SELFISH:VERSION:X.Y.Z -->` in CLAUDE.md. Then read `${CLAUDE_PLUGIN_ROOT}/package.json` to get the actual plugin version. Compare the two. | Block version ≥ plugin version | ⚠ Warning: SELFISH block is outdated (found {block_version}, current {plugin_version}). Fix: run `/selfish:init` to update |
 | No conflicting routing | Grep for conflicting agent patterns (`executor`, `deep-executor`, `debugger`, `code-reviewer`) outside SELFISH block that could intercept selfish intents | No conflicts or conflicts are inside other tool blocks | ⚠ Warning: found agent routing that may conflict with selfish skills. Review `~/.claude/CLAUDE.md` |
 
 ### Category 4: Pipeline State
@@ -68,9 +68,9 @@ Run ALL checks regardless of earlier failures. Do not short-circuit.
 | Check | How | Pass | Fail |
 |-------|-----|------|------|
 | No stale pipeline flag | Check `.claude/.selfish-active` | File does not exist (no active pipeline) | ⚠ Warning: stale pipeline flag found (feature: {name}). This may block normal operations. Fix: `rm .claude/.selfish-active .claude/.selfish-phase .claude/.selfish-ci-passed` or run `/selfish:resume` |
-| No orphaned artifacts | Glob `specs/*/spec.md` | No specs directories, or all are from active pipeline | ⚠ Warning: orphaned `specs/{name}/` found. Left over from a previous pipeline. Fix: `rm -rf specs/{name}/` |
+| No orphaned artifacts | Glob `.claude/selfish/specs/*/spec.md` | No specs directories, or all are from active pipeline | ⚠ Warning: orphaned `.claude/selfish/specs/{name}/` found. Left over from a previous pipeline. Fix: `rm -rf .claude/selfish/specs/{name}/` |
 | No lingering safety tags | `git tag -l 'selfish/pre-*'` | No tags, or tags match active pipeline | ⚠ Warning: lingering safety tag `selfish/pre-{x}` found. Fix: `git tag -d selfish/pre-{x}` |
-| Checkpoint state | Read `memory/checkpoint.md` if exists | No checkpoint (clean), or checkpoint is from current session | ⚠ Warning: stale checkpoint from {date}. Fix: run `/selfish:resume` to continue or delete `memory/checkpoint.md` |
+| Checkpoint state | Read `.claude/selfish/memory/checkpoint.md` if exists | No checkpoint (clean), or checkpoint is from current session | ⚠ Warning: stale checkpoint from {date}. Fix: run `/selfish:resume` to continue or delete `.claude/selfish/memory/checkpoint.md` |
 
 ### Category 5: Hook Health
 
@@ -149,7 +149,7 @@ Pipeline State
 
 Hook Health
   ✓ hooks.json valid
-  ✓ All 17 scripts exist
+  ✓ All scripts exist
   ✓ All scripts executable
 
 ─────────────────────────

package/commands/implement.md

@@ -38,7 +38,7 @@ git tag -f selfish/pre-implement
 ### 1. Load Context
 
 1. **Current branch** → `BRANCH_NAME`
-2. Load the following files from `specs/{feature}/`:
+2. Load the following files from `.claude/selfish/specs/{feature}/`:
    - **tasks.md** (required) — abort if missing: "tasks.md not found. Run `/selfish:tasks` first."
    - **plan.md** (required) — abort if missing
    - **spec.md** (for reference)
@@ -48,6 +48,10 @@ git tag -f selfish/pre-implement
    - Group by phase
    - Build dependency graph (validate DAG — abort if circular)
    - Identify already-completed `[x]` tasks
+4. **Recent changes**: run `git log --oneline -20` to understand what changed recently (context for implementation)
+5. **Smoke test**: run `{config.gate}` before starting implementation:
+   - If it fails → diagnose before implementing (existing code is broken — fix first or report to user)
+   - If it passes → baseline confirmed, proceed to implementation
 
 ### 2. Check Progress
 
@@ -145,7 +149,7 @@ When a worker agent exits with error (non-zero return or timeout):
 
 #### Phase Completion Gate (3 steps)
 
-> **Always** read `docs/phase-gate-protocol.md` first and perform the 3 steps (CI gate → Mini-Review → Auto-Checkpoint) in order.
+> **Always** read `${CLAUDE_PLUGIN_ROOT}/docs/phase-gate-protocol.md` first and perform the 3 steps (CI gate → Mini-Review → Auto-Checkpoint) in order.
 > Cannot advance to the next phase without passing the gate. Abort and report to user after 3 consecutive CI failures.
 
 After passing the gate, create a phase rollback point:

package/commands/init.md

@@ -188,6 +188,9 @@ User-only (not auto-triggered — inform user on request):
 - `selfish:doctor` — inform user when health check is requested
 - `selfish:architect` — inform user when architecture review is requested
 - `selfish:security` — inform user when security scan is requested
+- `selfish:checkpoint` — inform user when session save is requested
+- `selfish:resume` — inform user when session restore is requested
+- `selfish:principles` — inform user when project principles management is requested
 
 ## Pipeline
 

package/commands/plan.md

@@ -26,11 +26,11 @@ model: sonnet
 ### 1. Load Context
 
 1. Check **current branch** → `BRANCH_NAME`
-2. Find **specs/{feature}/spec.md**:
-   - Search under `specs/` for a directory matching the current branch name or `$ARGUMENTS`
+2. Find **.claude/selfish/specs/{feature}/spec.md**:
+   - Search under `.claude/selfish/specs/` for a directory matching the current branch name or `$ARGUMENTS`
    - If not found: print "spec.md not found. Run `/selfish:spec` first." then **abort**
 3. Read full **spec.md**
-4. Read **memory/principles.md** (if present)
+4. Read **.claude/selfish/memory/principles.md** (if present)
 5. Read **CLAUDE.md** project context
 
 ### 2. Clarification Check
@@ -39,7 +39,7 @@ model: sonnet
   - Warn user: "There are unresolved clarification items. Do you want to continue?"
   - If user chooses to stop → guide to `/selfish:clarify` then **abort**
 
-### 3. Phase 0 — Research (if needed)
+### 3. Phase 0 — Research (ReWOO pattern, if needed)
 
 Extract technical uncertainties from spec.md:
 
@@ -47,22 +47,40 @@ Extract technical uncertainties from spec.md:
 2. Are performance requirements unverified?
 3. Is the integration approach with the existing codebase unclear?
 
-**If there are uncertain items**:
-- Resolve each via WebSearch/codebase exploration
-- Record results in `specs/{feature}/research.md`:
-  ```markdown
-  ## {topic}
-  **Decision**: {chosen approach}
-  **Rationale**: {reason}
-  **Alternatives**: {other approaches considered}
-  **Source**: {URL or file path}
+**If no uncertain items**: skip Phase 0.
+
+**If there are uncertain items**, follow the 3-step ReWOO flow:
+
+#### Step 1: Plan (enumerate all topics — NO execution yet)
+List all research topics as a numbered list:
+```
+1. {topic1} — {what we need to know}
+2. {topic2} — {what we need to know}
+3. {topic3} — {what we need to know}
+```
+
+#### Step 2: Execute (parallel for independent topics)
+- If topics are independent (no result dependency): launch parallel Task() calls in a **single message**:
   ```
+  Task("Research: {topic1}", subagent_type: "general-purpose")
+  Task("Research: {topic2}", subagent_type: "general-purpose")
+  ```
+- If a topic depends on another's result: execute sequentially after the dependency resolves
+- For 1-2 topics: resolve directly via WebSearch/codebase exploration (no delegation needed)
 
-**If no uncertain items**: skip Phase 0.
+#### Step 3: Solve (consolidate all results)
+Collect all results and record in `.claude/selfish/specs/{feature}/research.md`:
+```markdown
+## {topic}
+**Decision**: {chosen approach}
+**Rationale**: {reason}
+**Alternatives**: {other approaches considered}
+**Source**: {URL or file path}
+```
 
 ### 4. Phase 1 — Write Design
 
-Create `specs/{feature}/plan.md`. **Must** follow the structure below:
+Create `.claude/selfish/specs/{feature}/plan.md`. **Must** follow the structure below:
 
 ```markdown
 # Implementation Plan: {feature name}
@@ -81,7 +99,7 @@ Create `specs/{feature}/plan.md`. **Must** follow the structure below:
 - **Constraints**: {constraints extracted from spec}
 
 ## Principles Check
-{if memory/principles.md exists: validation results against MUST principles}
+{if .claude/selfish/memory/principles.md exists: validation results against MUST principles}
 {if violations possible: state explicitly + justification}
 
 ## Architecture Decision
@@ -141,7 +159,7 @@ Create `specs/{feature}/plan.md`. **Must** follow the structure below:
 
 ### 5. Critic Loop
 
-> **Always** read `docs/critic-loop-rules.md` first and follow it.
+> **Always** read `${CLAUDE_PLUGIN_ROOT}/docs/critic-loop-rules.md` first and follow it.
 
 Run the critic loop until convergence. Safety cap: 7 passes.
 
@@ -150,7 +168,7 @@ Run the critic loop until convergence. Safety cap: 7 passes.
 | **COMPLETENESS** | Are all requirements (FR-*) from spec.md reflected in the plan? |
 | **FEASIBILITY** | Is it compatible with the existing codebase? Are dependencies available? |
 | **ARCHITECTURE** | Does it comply with {config.architecture} rules? |
-| **RISK** | Are there any unidentified risks? Additionally, if `memory/retrospectives/` directory contains files from previous pipeline runs, load each file and check whether the current plan addresses the patterns recorded there. Tag matched patterns with `[RETRO-CHECKED]`. |
+| **RISK** | Are there any unidentified risks? Additionally, if `.claude/selfish/memory/retrospectives/` directory contains files from previous pipeline runs, load each file and check whether the current plan addresses the patterns recorded there. Tag matched patterns with `[RETRO-CHECKED]`. |
 | **PRINCIPLES** | Does it not violate the MUST principles in principles.md? |
 
 **On FAIL**: auto-fix and continue to next pass.
@@ -159,22 +177,12 @@ Run the critic loop until convergence. Safety cap: 7 passes.
 **On CONVERGE**: `✓ Critic converged ({N} passes, {M} fixes, {E} escalations)`
 **On SAFETY CAP**: `⚠ Critic safety cap ({N} passes). Review recommended.`
 
-### 6. Agent Teams (if needed)
-
-If research items are 3 or more, delegate to parallel research agents via Task tool:
-
-```
-Task("Research: {topic1}", subagent_type: "general-purpose")
-Task("Research: {topic2}", subagent_type: "general-purpose")
-→ collect results → consolidate into research.md
-```
-
-### 7. Final Output
+### 6. Final Output
 
 ```
 Plan generated
-├─ specs/{feature}/plan.md
-├─ specs/{feature}/research.md (if research was performed)
+├─ .claude/selfish/specs/{feature}/plan.md
+├─ .claude/selfish/specs/{feature}/research.md (if research was performed)
 ├─ Critic: converged ({N} passes, {M} fixes, {E} escalations)
 └─ Next step: /selfish:tasks
 ```

package/commands/principles.md

@@ -14,7 +14,7 @@ model: haiku
 # /selfish:principles — Manage Project Principles
 
 > Creates and manages the project's core principles (constitution).
-> Stored in memory/principles.md and referenced across all sessions.
+> Stored in .claude/selfish/memory/principles.md and referenced across all sessions.
 
 ## Arguments
 
@@ -32,7 +32,7 @@ model: haiku
 
 ### 1. Check Current State
 
-Read `memory/principles.md`:
+Read `.claude/selfish/memory/principles.md`:
 - If present: load existing principles
 - If absent: empty state (show `init` instructions)
 
@@ -101,7 +101,7 @@ Collect principles interactively:
 
 ## Notes
 
-- **Persistent storage**: Saved to memory/principles.md and maintained across sessions.
+- **Persistent storage**: Saved to .claude/selfish/memory/principles.md and maintained across sessions.
 - **Auto-referenced**: Automatically loaded and validated by /selfish:plan and /selfish:architect.
 - **Keep it concise**: Maintain no more than 10 principles. Too many reduces effectiveness.
 - **Avoid duplication with CLAUDE.md**: Do not re-register rules already present in CLAUDE.md as principles.

package/commands/research.md

@@ -16,7 +16,7 @@ model: sonnet
 # /selfish:research — Technical Research
 
 > Investigates technical questions and summarizes conclusions.
-> Results are persisted to memory/research/{topic}.md.
+> Results are persisted to .claude/selfish/memory/research/{topic}.md.
 
 ## Arguments
 
@@ -33,7 +33,7 @@ Extract from `$ARGUMENTS`:
 
 ### 2. Check Existing Research
 
-Check `memory/research/` directory for related prior research:
+Check `.claude/selfish/memory/research/` directory for related prior research:
 - If found: load existing content and decide whether an update is needed
 - If not found: proceed with new research
 
@@ -88,7 +88,7 @@ Source priority:
 
 ### 5. Save
 
-- Save to `memory/research/{topic-kebab-case}.md`
+- Save to `.claude/selfish/memory/research/{topic-kebab-case}.md`
 - If the file already exists, update it (refresh the date)
 
 ### 6. Final Output
@@ -96,7 +96,7 @@ Source priority:
 ```
 Research complete
 ├─ Topic: {topic}
-├─ Saved: memory/research/{filename}.md
+├─ Saved: .claude/selfish/memory/research/{filename}.md
 ├─ Conclusion: {one-line summary}
 └─ Sources: {number of key sources}
 ```
@@ -106,4 +106,4 @@ Research complete
 - **Current date basis**: Use WebSearch to verify up-to-date information rather than relying on knowledge cutoff.
 - **Sources required**: Cite sources for all technical claims.
 - **Project context**: Derive conclusions applicable to this project, not generic research.
-- **Persistent storage**: Save to memory/research/ for reuse across sessions.
+- **Persistent storage**: Save to .claude/selfish/memory/research/ for reuse across sessions.

package/commands/resume.md

@@ -12,7 +12,7 @@ allowed-tools:
 
 # /selfish:resume — Restore Session
 
-> Restores the previous session state from memory/checkpoint.md and resumes work.
+> Restores the previous session state from .claude/selfish/memory/checkpoint.md and resumes work.
 
 ## Arguments
 
@@ -22,7 +22,7 @@ allowed-tools:
 
 ### 1. Load Checkpoint
 
-Read `memory/checkpoint.md`:
+Read `.claude/selfish/memory/checkpoint.md`:
 - If not found: output "No saved checkpoint found." then **stop**
 - If found: parse the full contents
 
@@ -34,7 +34,7 @@ Compare the checkpoint state against the current environment:
    - If different: warn + suggest switching
 2. **File state**: Have any files changed since the checkpoint?
    - Check for new commits with `git log {checkpoint hash}..HEAD --oneline`
-3. **Feature directory**: Does specs/{feature}/ still exist?
+3. **Feature directory**: Does .claude/selfish/specs/{feature}/ still exist?
 
 ### 3. Report State
 

package/commands/review.md

@@ -1,9 +1,10 @@
 ---
 name: selfish:review
-description: "Code review (read-only)"
+description: "Code review"
 argument-hint: "[scope: file path, PR number, or staged]"
 allowed-tools:
   - Read
+  - Write
   - Grep
   - Glob
   - Bash
@@ -133,13 +134,13 @@ For each changed file, examine from the following perspectives:
 
 ### 5. Retrospective Check
 
-If `memory/retrospectives/` directory exists, load retrospective files and check:
+If `.claude/selfish/memory/retrospectives/` directory exists, load retrospective files and check:
 - Were there recurring Critical finding categories in past reviews? Prioritize those perspectives.
 - Were there false positives that wasted effort? Reduce sensitivity for those patterns.
 
 ### 6. Critic Loop
 
-> **Always** read `docs/critic-loop-rules.md` first and follow it.
+> **Always** read `${CLAUDE_PLUGIN_ROOT}/docs/critic-loop-rules.md` first and follow it.
 
 Run the critic loop until convergence. Safety cap: 5 passes.
 
@@ -154,11 +155,26 @@ Run the critic loop until convergence. Safety cap: 5 passes.
 **On CONVERGE**: `✓ Critic converged ({N} passes, {M} fixes, {E} escalations)`
 **On SAFETY CAP**: `⚠ Critic safety cap ({N} passes). Review recommended.`
 
-### 7. Archive Review Report
+### 7. Retrospective Entry (if new pattern found)
 
-When running inside a pipeline (specs/{feature}/ exists), persist the review results:
+If this review reveals a recurring pattern not previously documented in `.claude/selfish/memory/retrospectives/`:
 
-1. Write full review output (Summary table + Detailed Findings + Positives) to `specs/{feature}/review-report.md`
+Append to `.claude/selfish/memory/retrospectives/{YYYY-MM-DD}.md`:
+```markdown
+## Pattern: {category}
+**What happened**: {concrete description}
+**Root cause**: {why this keeps occurring}
+**Prevention rule**: {actionable rule — usable in future plan/implement phases}
+**Severity**: Critical | Warning
+```
+
+Only write if the pattern is new and actionable. Generic observations are prohibited.
+
+### 8. Archive Review Report
+
+When running inside a pipeline (.claude/selfish/specs/{feature}/ exists), persist the review results:
+
+1. Write full review output (Summary table + Detailed Findings + Positives) to `.claude/selfish/specs/{feature}/review-report.md`
 2. Include metadata header:
    ```markdown
    # Review Report: {feature name}
@@ -166,11 +182,11 @@ When running inside a pipeline (specs/{feature}/ exists), persist the review res
    > Files reviewed: {count}
    > Findings: Critical {N} / Warning {N} / Info {N}
    ```
-3. This file survives Clean phase (copied to `memory/reviews/{feature}-{date}.md` before specs/ deletion)
+3. This file survives Clean phase (copied to `.claude/selfish/memory/reviews/{feature}-{date}.md` before .claude/selfish/specs/ deletion)
 
 When running standalone (no active pipeline), skip archiving — display results in console only.
 
-### 8. Final Output
+### 9. Final Output
 
 ```
 Review complete

package/commands/spec.md

@@ -29,7 +29,7 @@ model: sonnet
 2. Determine **feature name**:
    - Extract 2-3 key keywords from `$ARGUMENTS`
    - Convert to kebab-case (e.g., "add user authentication" → `user-auth`)
-3. **Create directory**: `specs/{feature-name}/` (create parent `specs/` directory if it does not exist)
+3. **Create directory**: `.claude/selfish/specs/{feature-name}/` (create parent `.claude/selfish/specs/` directory if it does not exist)
 4. If already exists, confirm with user: "Overwrite existing spec?"
 
 ### 2. Explore Codebase
@@ -42,7 +42,7 @@ Before writing the spec, understand the current project structure:
 
 ### 3. Write Spec
 
-Create `specs/{feature-name}/spec.md`:
+Create `.claude/selfish/specs/{feature-name}/spec.md`:
 
 ```markdown
 # Feature Spec: {feature name}
@@ -78,7 +78,7 @@ Create `specs/{feature-name}/spec.md`:
 - **NFR-001**: {performance/security/accessibility etc.}
 
 ### Auto-Suggested NFRs
-{Load `docs/nfr-templates.md` and select 3-5 relevant NFRs based on the project type detected from selfish.config.md}
+{Load `${CLAUDE_PLUGIN_ROOT}/docs/nfr-templates.md` and select 3-5 relevant NFRs based on the project type detected from selfish.config.md}
 - **NFR-A01** [AUTO-SUGGESTED]: {suggestion from matching project type template}
 - **NFR-A02** [AUTO-SUGGESTED]: {suggestion}
 - **NFR-A03** [AUTO-SUGGESTED]: {suggestion}
@@ -106,13 +106,13 @@ Create `specs/{feature-name}/spec.md`:
 
 ### 4. Retrospective Check
 
-If `memory/retrospectives/` directory exists, load retrospective files and check:
+If `.claude/selfish/memory/retrospectives/` directory exists, load retrospective files and check:
 - Were there previous `[AUTO-RESOLVED]` items that turned out wrong? Flag similar patterns.
 - Were there scope-related issues in past specs? Warn about similar ambiguities.
 
 ### 5. Critic Loop
 
-> **Always** read `docs/critic-loop-rules.md` first and follow it.
+> **Always** read `${CLAUDE_PLUGIN_ROOT}/docs/critic-loop-rules.md` first and follow it.
 
 Run the critic loop until convergence. Safety cap: 5 passes.
 
@@ -133,7 +133,7 @@ Run the critic loop until convergence. Safety cap: 5 passes.
 
 ```
 Spec generated
-├─ specs/{feature-name}/spec.md
+├─ .claude/selfish/specs/{feature-name}/spec.md
 ├─ User Stories: {count}
 ├─ Requirements: FR {count}, NFR {count}
 ├─ Unresolved: {[NEEDS CLARIFICATION] count}

package/commands/tasks.md

@@ -22,7 +22,7 @@ model: sonnet
 
 ### 1. Load Context
 
-1. Load from `specs/{feature}/`:
+1. Load from `.claude/selfish/specs/{feature}/`:
    - **plan.md** (required) — stop if missing: "Run /selfish:plan first."
    - **spec.md** (required)
    - **research.md** (if present)
@@ -73,20 +73,20 @@ Decompose tasks per Phase defined in plan.md.
 1. **1 task = 1 file** principle (where possible)
 2. **Same file = sequential**, **different files = [P] candidate**
 3. **Explicit dependencies**: Use `depends: [T001, T002]` to declare blocking dependencies. Tasks without `depends:` and with [P] marker are immediately parallelizable.
-4. **[P] physical validation**: Before finalizing tasks.md, run `"${CLAUDE_PLUGIN_ROOT}/scripts/selfish-parallel-validate.sh" specs/{feature}/tasks.md` to verify no file path overlaps exist among [P] tasks within the same phase. Fix any conflicts before proceeding.
+4. **[P] physical validation**: Before finalizing tasks.md, run `"${CLAUDE_PLUGIN_ROOT}/scripts/selfish-parallel-validate.sh" .claude/selfish/specs/{feature}/tasks.md` to verify no file path overlaps exist among [P] tasks within the same phase. Fix any conflicts before proceeding.
 5. **Dependency graph must be a DAG**: no circular dependencies allowed. Validate before output.
 6. **Test tasks**: Include a verification task for each testable unit
 7. **Phase gate**: Add a `{config.gate}` validation task at the end of each Phase
 
 ### 3. Retrospective Check
 
-If `memory/retrospectives/` directory exists, load retrospective files and check:
+If `.claude/selfish/memory/retrospectives/` directory exists, load retrospective files and check:
 - Were there previous parallel conflict issues ([P] file overlaps)? Flag similar file patterns.
 - Were there tasks that were over-decomposed or under-decomposed? Adjust granularity.
 
 ### 4. Critic Loop
 
-> **Always** read `docs/critic-loop-rules.md` first and follow it.
+> **Always** read `${CLAUDE_PLUGIN_ROOT}/docs/critic-loop-rules.md` first and follow it.
 
 Run the critic loop until convergence. Safety cap: 5 passes.
 
@@ -116,11 +116,11 @@ Every FR-*/NFR-* must be mapped to at least one task.
 
 ### 6. Final Output
 
-Save to `specs/{feature}/tasks.md`, then:
+Save to `.claude/selfish/specs/{feature}/tasks.md`, then:
 
 ```
 Tasks generated
-├─ specs/{feature}/tasks.md
+├─ .claude/selfish/specs/{feature}/tasks.md
 ├─ Tasks: {total count} ({[P] count} parallelizable)
 ├─ Phases: {phase count}
 ├─ Coverage: FR {coverage}%, NFR {coverage}%

package/commands/test.md

@@ -75,7 +75,7 @@ Confirm strategy with user before proceeding.
 
 ### 4. Critic Loop
 
-> **Always** read `docs/critic-loop-rules.md` first and follow it.
+> **Always** read `${CLAUDE_PLUGIN_ROOT}/docs/critic-loop-rules.md` first and follow it.
 
 Run the critic loop until convergence. Safety cap: 5 passes.
 

package/docs/critic-loop-rules.md

@@ -6,7 +6,11 @@
 
 1. **Minimum findings**: In each Critic round, **at least 1 concern, improvement point, or verification rationale per criterion** must be stated. If there are no issues, explain specifically "why there are no issues."
 2. **Checklist responses**: For each criterion, output takes the form of answering specific questions. Single-word "PASS" is prohibited.
-3. **Adversarial Pass**: At the end of every round, **"1 scenario in which this output fails"** must be stated. If the scenario is realistic, convert to FAIL and fix.
+3. **Adversarial Pass**: At the end of every round, apply **3 progressive critic perspectives**:
+   - **Skeptic**: "Which assumption here is most likely wrong?"
+   - **Devil's Advocate**: "How could this design be misused or fail unexpectedly?"
+   - **Edge-case Hunter**: "What input would cause this to fail silently?"
+   State one failure scenario per perspective. If any scenario is realistic → convert to FAIL and fix. If all are unrealistic → state quantitative rationale for each.
 4. **Quantitative rationale**: Instead of qualitative judgments like "none" or "compliant," present quantitative data such as "M of N confirmed," "Y of X lines applicable."
 
 ## Verdict System (4 types)
@@ -75,8 +79,11 @@ Options:
 [Criterion1] {question} → {answer + quantitative rationale}
   Verdict: PASS | FAIL ({fix description}) | ESCALATE | DEFER ({reason})
 [Criterion2] ...
-[ADVERSARIAL] Failure scenario: {specific scenario}
-  → Realistic? {Y → FAIL + fix / N → state rationale}
+[ADVERSARIAL]
+  Skeptic: {assumption most likely wrong}
+  Devil's Advocate: {misuse or unexpected failure}
+  Edge-case Hunter: {input causing silent failure}
+  → Realistic? {Y → FAIL + fix / N → state quantitative rationale per perspective}
 === Result: {CONVERGED ({N} passes, {M} fixes, {E} escalations) | CONTINUE ({N} fixed) | ESCALATE ({N} items) | SAFETY CAP} ===
 ```
 

package/docs/phase-gate-protocol.md

@@ -34,7 +34,7 @@ Quantitatively inspect `{config.mini_review}` items for files changed within the
 After passing the Phase gate, automatically save session state:
 
 ```markdown
-# memory/checkpoint.md auto-update
+# .claude/selfish/memory/checkpoint.md auto-update
 Current Phase: {N}/{total}
 Completed tasks: {list of completed IDs}
 Changed files: {file list}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "selfish-pipeline",
-  "version": "1.2.1",
+  "version": "1.3.0",
   "description": "Claude Code plugin that automates the full dev cycle — spec, plan, tasks, implement, review, clean. 18 commands, 15 hooks, 5 presets.",
   "bin": {
     "selfish-pipeline": "./bin/cli.mjs"

package/scripts/pre-compact-checkpoint.sh

@@ -59,8 +59,8 @@ fi
 # Check tasks.md progress status
 TASKS_DONE=0
 TASKS_TOTAL=0
-if [ -n "$PIPELINE_FEATURE" ] && [ -d "$PROJECT_DIR/specs/$PIPELINE_FEATURE" ]; then
-  TASKS_FILE="$PROJECT_DIR/specs/$PIPELINE_FEATURE/tasks.md"
+if [ -n "$PIPELINE_FEATURE" ] && [ -d "$PROJECT_DIR/.claude/selfish/specs/$PIPELINE_FEATURE" ]; then
+  TASKS_FILE="$PROJECT_DIR/.claude/selfish/specs/$PIPELINE_FEATURE/tasks.md"
   if [ -f "$TASKS_FILE" ]; then
     TASKS_DONE=$(grep -cE '\[x\]' "$TASKS_FILE" 2>/dev/null || echo 0)
     TASKS_TOTAL=$(grep -cE '\[(x| )\]' "$TASKS_FILE" 2>/dev/null || echo 0)
@@ -107,6 +107,6 @@ $STAGED_LIST
 EOF
 
 # Inject context via stdout (Claude can see this info after compaction)
-echo "Auto-checkpoint saved to memory/checkpoint.md (branch: $BRANCH, pipeline: ${PIPELINE_FEATURE:-inactive})"
+echo "Auto-checkpoint saved to .claude/selfish/memory/checkpoint.md (branch: $BRANCH, pipeline: ${PIPELINE_FEATURE:-inactive})"
 
 exit 0

package/scripts/selfish-preflight-check.sh

@@ -14,33 +14,122 @@ WARN_COUNT=0
 FAIL_COUNT=0
 
 CI_COMMAND=""
+CI_SOURCE=""
+
+# ── Helpers ──────────────────────────────────────────────
+
+# shellcheck disable=SC2317
+_detect_pm() {
+  local dir="$1"
+  if [[ -f "$dir/pnpm-lock.yaml" ]]; then printf 'pnpm'
+  elif [[ -f "$dir/yarn.lock" ]]; then printf 'yarn'
+  elif [[ -f "$dir/bun.lock" || -f "$dir/bun.lockb" ]]; then printf 'bun'
+  elif [[ -f "$dir/package-lock.json" ]]; then printf 'npm'
+  elif [[ -f "$dir/package.json" ]] && command -v jq > /dev/null 2>&1; then
+    local pm_field
+    pm_field=$(jq -r '.packageManager // empty' "$dir/package.json" 2>/dev/null)
+    if [[ -n "$pm_field" ]]; then
+      printf '%s' "${pm_field%%@*}"
+    else
+      printf 'npm'
+    fi
+  else
+    printf 'npm'
+  fi
+}
+
+# shellcheck disable=SC2317
+_has_script() {
+  local pkg="$1" script="$2"
+  if command -v jq > /dev/null 2>&1; then
+    jq -e ".scripts[\"$script\"]" "$pkg" > /dev/null 2>&1
+  else
+    grep -qE "\"$script\"[[:space:]]*:" "$pkg" 2>/dev/null
+  fi
+}
+
+# ── PM Detection (lockfile-first, Turborepo pattern) ─────
+
+PM=$(_detect_pm "$PROJECT_DIR")
 
 printf 'Preflight Check:\n'
 
-# Check 1: CI command exists
-if [[ -f "$PROJECT_DIR/package.json" ]]; then
-  if jq -e '.scripts["test:all"]' "$PROJECT_DIR/package.json" > /dev/null 2>&1; then
-    CI_COMMAND="npm run test:all"
-  elif jq -e '.scripts["test"]' "$PROJECT_DIR/package.json" > /dev/null 2>&1; then
-    CI_COMMAND="npm test"
+# ── Check 1: CI command (4-tier cascade) ─────────────────
+
+# 1a. selfish.config.md — most explicit, user-configured
+CONFIG_FILE="$PROJECT_DIR/.claude/selfish.config.md"
+if [[ -z "$CI_COMMAND" && -f "$CONFIG_FILE" ]]; then
+  CI_COMMAND=$(grep -E '^\s*ci:\s*"[^"]*"' "$CONFIG_FILE" 2>/dev/null | head -1 | sed 's/.*ci: *"\([^"]*\)".*/\1/' || true)
+  if [[ -z "$CI_COMMAND" ]]; then
+    CI_COMMAND=$(grep -E '^\s*gate:\s*"[^"]*"' "$CONFIG_FILE" 2>/dev/null | head -1 | sed 's/.*gate: *"\([^"]*\)".*/\1/' || true)
   fi
+  [[ -n "$CI_COMMAND" ]] && CI_SOURCE="selfish.config.md"
 fi
 
+# 1b. Monorepo tools (turbo, nx, pnpm workspaces)
+if [[ -z "$CI_COMMAND" ]]; then
+  if [[ -f "$PROJECT_DIR/turbo.json" ]]; then
+    if [[ "$PM" == "pnpm" || "$PM" == "yarn" || "$PM" == "bun" ]]; then
+      CI_COMMAND="$PM turbo test"
+    else
+      CI_COMMAND="npx turbo test"
+    fi
+    CI_SOURCE="turbo.json"
+  elif [[ -f "$PROJECT_DIR/nx.json" ]]; then
+    CI_COMMAND="npx nx run-many --target=test"
+    CI_SOURCE="nx.json"
+  elif [[ -f "$PROJECT_DIR/pnpm-workspace.yaml" ]] && ! _has_script "$PROJECT_DIR/package.json" "test" 2>/dev/null; then
+    CI_COMMAND="pnpm -r test"
+    CI_SOURCE="pnpm-workspace.yaml"
+  fi
+fi
+
+# 1c. package.json scripts (PM-aware)
+if [[ -z "$CI_COMMAND" && -f "$PROJECT_DIR/package.json" ]]; then
+  for script in "test:all" "test" "ci" "check"; do
+    if _has_script "$PROJECT_DIR/package.json" "$script"; then
+      if [[ "$script" == "test" ]]; then
+        CI_COMMAND="$PM test"
+      else
+        CI_COMMAND="$PM run $script"
+      fi
+      CI_SOURCE="package.json"
+      break
+    fi
+  done
+fi
+
+# 1d. Makefile
+if [[ -z "$CI_COMMAND" && -f "$PROJECT_DIR/Makefile" ]]; then
+  MAKE_TARGET=$(grep -E '^(test|check|ci):' "$PROJECT_DIR/Makefile" 2>/dev/null | head -1 | cut -d: -f1 || true)
+  if [[ -n "$MAKE_TARGET" ]]; then
+    CI_COMMAND="make $MAKE_TARGET"
+    CI_SOURCE="Makefile"
+  fi
+fi
+
+# Report CI command result
 if [[ -n "$CI_COMMAND" ]]; then
-  printf '  \xe2\x9c\x93 CI command: %s\n' "$CI_COMMAND"
+  printf '  \xe2\x9c\x93 CI command: %s  [%s]\n' "$CI_COMMAND" "$CI_SOURCE"
   PASS_COUNT=$((PASS_COUNT + 1))
-else
+elif [[ -f "$PROJECT_DIR/package.json" ]]; then
+  # package.json exists but no CI script — configuration error
   printf '  \xe2\x9c\x97 CI command: no test script found in package.json\n'
   FAIL_COUNT=$((FAIL_COUNT + 1))
+else
+  # No recognizable ecosystem — warn instead of fail
+  printf '  \xe2\x9a\xa0 CI command: not detected (configure ci: in selfish.config.md)\n'
+  WARN_COUNT=$((WARN_COUNT + 1))
 fi
 
-# Check 2: Dependencies installed
+# ── Check 2: Dependencies installed ──────────────────────
+
 if [[ -f "$PROJECT_DIR/package.json" ]]; then
   if [[ -d "$PROJECT_DIR/node_modules" ]]; then
     printf '  \xe2\x9c\x93 Dependencies: node_modules present\n'
     PASS_COUNT=$((PASS_COUNT + 1))
   else
-    printf '  \xe2\x9a\xa0 Dependencies: node_modules not found (run npm install)\n'
+    printf '  \xe2\x9a\xa0 Dependencies: node_modules not found (run %s install)\n' "$PM"
     WARN_COUNT=$((WARN_COUNT + 1))
   fi
 else
@@ -48,7 +137,8 @@ else
   PASS_COUNT=$((PASS_COUNT + 1))
 fi
 
-# Check 3: Shellcheck available
+# ── Check 3: Shellcheck available ────────────────────────
+
 if command -v shellcheck > /dev/null 2>&1; then
   printf '  \xe2\x9c\x93 Shellcheck: installed\n'
   PASS_COUNT=$((PASS_COUNT + 1))
@@ -57,7 +147,8 @@ else
   WARN_COUNT=$((WARN_COUNT + 1))
 fi
 
-# Check 4: Git state
+# ── Check 4: Git state ───────────────────────────────────
+
 if git -C "$PROJECT_DIR" rev-parse --is-inside-work-tree > /dev/null 2>&1; then
   DIRTY_COUNT=$(git -C "$PROJECT_DIR" status --porcelain 2>/dev/null | wc -l | tr -d ' ')
   if [[ "$DIRTY_COUNT" -eq 0 ]]; then
@@ -72,7 +163,8 @@ else
   WARN_COUNT=$((WARN_COUNT + 1))
 fi
 
-# Check 5: No active pipeline
+# ── Check 5: No active pipeline ──────────────────────────
+
 ACTIVE_FILE="$PROJECT_DIR/.claude/.selfish-active"
 if [[ -f "$ACTIVE_FILE" ]]; then
   ACTIVE_NAME=$(head -1 "$ACTIVE_FILE" 2>/dev/null | tr -d '\n\r' | cut -c1-100 || printf 'unknown')
@@ -83,7 +175,8 @@ else
   PASS_COUNT=$((PASS_COUNT + 1))
 fi
 
-# Result
+# ── Result ────────────────────────────────────────────────
+
 printf '\n'
 if [[ "$FAIL_COUNT" -gt 0 ]]; then
   if [[ "$WARN_COUNT" -gt 0 ]]; then

package/scripts/session-start-context.sh

@@ -36,7 +36,7 @@ if [ -f "$PIPELINE_FLAG" ]; then
   OUTPUT="[SELFISH PIPELINE ACTIVE] Feature: $FEATURE"
 
   # tasks.md progress
-  TASKS_FILE="$PROJECT_DIR/specs/$FEATURE/tasks.md"
+  TASKS_FILE="$PROJECT_DIR/.claude/selfish/specs/$FEATURE/tasks.md"
   if [ -f "$TASKS_FILE" ]; then
     DONE=$(grep -cE '\[x\]' "$TASKS_FILE" 2>/dev/null || echo 0)
     TOTAL=$(grep -cE '\[(x| )\]' "$TASKS_FILE" 2>/dev/null || echo 0)