oh-my-customcodex 0.4.11 → 0.4.13

package/templates/.claude/skills/professor-triage/SKILL.md

@@ -1,323 +1,90 @@
 ---
 name: professor-triage
-description: Analyze GitHub issues against current codebase and perform automated triage with priority assessment
+description: Analyze GitHub issues against the current codebase and perform automated triage with priority assessment
 scope: harness
-version: 2.2.0
+version: 2.4.0
 user-invocable: true
 effort: high
 context: fork
 argument-hint: "[issue-numbers...] [--label <label>] [--state <state>] [--since <date>]"
 ---
 
-# /professor-triage — Codebase-Driven Issue Triage
+# /professor-triage - Codebase-Driven Issue Triage
 
 ## Purpose
 
-Analyzes GitHub issues directly against the current codebase. For each issue, searches relevant code, assesses impact and blast radius, determines whether the issue has already been resolved, and performs automated triage with priority and size estimation. Produces a cross-analysis report and executes low-risk triage actions automatically.
+Analyze GitHub issues against the current codebase. For each issue, search relevant code, assess impact and blast radius, determine whether the issue has already been resolved, and perform low-risk triage actions with evidence.
 
-## Usage
-
-```
-/professor-triage                          # Default: --state open (excludes verify-done)
-/professor-triage 587 589 590 591 592      # Direct issue numbers
-/professor-triage --label codex-release    # Custom label filter
-/professor-triage --since 2026-03-20       # Date filter
-```
-
-## Workflow
+**Full phase detail**: `guides/professor-triage/phases.md`
+**Checklist**: `guides/professor-triage/checklists.md`
 
-### Phase 1: Gather
-
-1. Parse arguments to determine target issues:
-   - If issue numbers provided: use those directly
-   - If `--label` provided: `gh issue list --label <label> --state <state> --json number`
-   - Default: `gh issue list --state open --json number` + exclude issues with `verify-done` label
-   - If `--since` provided: add `--search "created:>YYYY-MM-DD"` filter
+## Usage
 
-2. For each issue, fetch full details:
 ```bash
-gh issue view NNN --json number,title,body,comments,labels,createdAt
-```
-
-3. For batches >20 issues, prefer `gh api graphql` for batch fetching to respect GitHub API rate limits (5000/hour authenticated).
-
-4. If filter returns 0 results: if `--label` was used, check label existence via `gh label list`. Report if label missing. If default filter, report "No open issues without verify-done label found."
-
-### Phase 2: Codebase Analysis
-
-For each issue, perform direct codebase analysis:
-
-**2A: Context Extraction** — From issue title and body, extract:
-- File paths mentioned (regex: backtick-wrapped paths, `:\d+` line refs, `(L\d+)`, `(lines \d+-\d+)`)
-- Error messages or stack traces
-- Keywords (function names, class names, config keys, module names)
-- Component areas mentioned (e.g., "auth", "CI", "hooks")
-
-**2B: Codebase Search** — Delegate to Explore agent(s):
-- Search for extracted keywords using Grep across the codebase
-- Find related files using Glob patterns derived from keywords
-- For explicitly mentioned files, verify existence and read relevant sections
-- For error messages, trace to source location
-- Map import/dependency relationships for affected files
-
-**2C: Impact Assessment** — For each relevant file found:
-- Read current state of the code
-- Check recent changes: `git log --since=<issue_created_date> --oneline -- <file>`
-- Determine if the issue has already been addressed by recent commits
-- Assess blast radius (what depends on this code, what does this code depend on)
-
-**2D: Structured Finding** — Produce per-issue analysis:
-
-| Field | Content |
-|-------|---------|
-| Affected files | List with status: `exists` ✅ / `missing` ❌ / `changed-since-issue` ⚠️ |
-| Architecture impact | Breaking changes, dependency effects, scope of change |
-| Implementation path | Concrete steps with file:line references from current codebase |
-| Risk level | P1 (critical/security/breaking) / P2 (moderate/compat) / P3 (nice-to-have) |
-| Size estimate | XS (<1h) / S (1-3h) / M (3-8h) / L (1-3d) / XL (>3d) |
-| Already resolved? | Yes / No / Partial — with git evidence (commit hash, PR number) |
-
-**Parallelization (R009/R018):**
-- 1-3 issues → single Explore agent per issue (parallel per R009)
-- 4-10 issues → parallel Explore agents, max 4 concurrent (R009)
-- 10+ issues or 3+ Explore agents needed → Agent Teams per R018
-
-**Delegation**: All codebase search delegated to Explore agent(s) with `model: haiku`. Orchestrator collects and synthesizes results.
-
-### Phase 3: Cross-Analyze
-
-**R010 note**: This is a read-only analytical step — no file writes. Per R010 exception, the orchestrator may perform this directly. For batches >15 issues, delegate to a dedicated cross-analysis agent with model: opus.
-
-Perform deep cross-analysis with full context from all issues:
-
-1. **Common patterns** — Identify findings that appear across multiple issues (e.g., same file referenced, same recommendation theme)
-2. **Duplicate/merge candidates** — Detect issues tracking the same underlying change:
-   - Same release series (e.g., alpha.3/5/6)
-   - Same upstream dependency
-   - Same affected component
-3. **Conflicting findings** — Where findings disagree across issues, resolve based on:
-   - Codebase evidence (Phase 2 results)
-   - Specificity (concrete code-level finding > abstract observation)
-   - Recency (newer findings > older ones)
-4. **Priority matrix** — Unified priority ranking:
-   - P1: Breaking changes, security issues, blocking bugs
-   - P2: Documentation gaps, compatibility updates, medium-risk items
-   - P3: Nice-to-have improvements, future considerations
-5. **Action determination** — Per-issue decision:
-   - `Close (Already Resolved)`: Phase 2 found issue already fixed by recent commits
-   - `Close (Not Applicable)`: Issue is irrelevant (internal dependency tag, no impact)
-   - `Close (Duplicate of #NNN)`: Superseded by another issue in the batch
-   - `Open — action required`: Real work needed
-   - `Open — monitoring`: Waiting for external trigger (e.g., stable release)
-   - `New issue needed`: Cross-analysis discovered issue not yet tracked
-
-### Phase 4: Multi-Perspective Analysis & Output
-
-For each analyzed issue, generate multi-perspective analysis comments and artifacts.
-
-**Parallelization (R009):**
-- Phase 4A + 4B: parallel (independent perspectives)
-- Phase 4C: after 4A + 4B complete (synthesis requires both inputs)
-- Phase 4D + 4E: parallel (independent outputs, both depend on 4C)
-- Phase 4F: after all above (verification gate)
-
-**4A: 🏛️ Senior Architect Analysis** — Delegate to arch-documenter (model: sonnet) to post GitHub comment:
-
+/professor-triage
+/professor-triage 587 589 590
+/professor-triage --label codex-release
+/professor-triage --since 2026-03-20
 ```
-## 🏛️ Senior Architect Analysis
 
-### 아키텍처 영향
-| 컴포넌트 | 영향 | 위험도 |
-|----------|------|--------|
-| {컴포넌트} | {설명} | {High/Medium/Low} |
+## Workflow Contract
 
-### 코드 수준 분석
-{Phase 2 코드베이스 분석의 구체적 file:line 참조}
+| Phase | Name | Owner | Notes |
+|-------|------|-------|-------|
+| 1 | Gather | Orchestrator | Fetch issue metadata and labels |
+| 2 | Codebase Analysis | Explore agents | Parallel code search and evidence mapping |
+| 3 | Cross-Analyze | Orchestrator or synthesis agent | Duplicate, priority, and action decisions |
+| 4 | Multi-Perspective Output | delegated analysis agents | Architect, colleague, professor synthesis, artifact |
+| 5 | Act | mgr-gitnerd | Labels, comments, close decisions |
 
-### 전략적 평가
-- **실현 가능성**: {근거가 포함된 평가}
-- **우선순위 권장**: {P1/P2/P3 및 근거}
+## Delegation Contract
 
-### 리스크 및 고려사항
-| 리스크 | 가능성 | 완화 방안 |
-|--------|--------|----------|
-| {리스크} | {High/Medium/Low} | {완화 방안} |
+| Phase | Agent | Mode |
+|-------|-------|------|
+| Phase 2 codebase search | Explore | bypassPermissions |
+| Phase 4A Senior Architect | general-purpose | bypassPermissions |
+| Phase 4B Project Colleague | general-purpose | bypassPermissions |
+| Phase 4C Professor Synthesis | general-purpose | bypassPermissions |
+| Phase 4D triage comment | mgr-gitnerd | bypassPermissions |
+| Phase 4E artifact report | general-purpose | bypassPermissions |
+| Phase 5 GitHub actions | mgr-gitnerd | bypassPermissions |
 
-**예상 작업량**: {XS/S/M/L/XL}
+Agent selection constraint: artifact-writing delegated agents need Bash access for compatibility-path bypasses. Do not use an agent whose tool policy prevents the required artifact protocol.
 
----
-_🏛️ Senior Architect perspective — `/professor-triage` v2.2.0_
-```
-
-**4B: 🤝 Project Colleague Review** — Delegate to arch-documenter (model: sonnet) to post GitHub comment:
+## Parallelization
 
-```
-## 🤝 Project Colleague Review
+- 1-3 issues: one Explore agent per issue, parallel when useful.
+- 4-10 issues: parallel Explore agents, max 4 concurrent.
+- 10+ issues: prefer a coordinated team surface when available.
+- Phase 4A and 4B are parallel; Phase 4C waits for both; Phase 4D and 4E are parallel after synthesis.
 
-### 구현 아이디어
-{구체적 코드 위치 및 file:line 참조가 포함된 변경 제안}
+## Sensitive-path compatibility note
 
-### 놓치기 쉬운 세부사항
-- {이름 충돌, 유효성 검사 우회, 경쟁 조건, 엣지 케이스}
+Codex-native artifacts go under `.codex/outputs/sessions/YYYY-MM-DD/professor-triage-HHmmss.md` and may be written with the repo's normal file-write APIs.
 
-### 권장 다음 단계
-1. {구체적 file/function 참조가 포함된 실행 가능한 단계}
-2. {실행 가능한 단계}
-3. {실행 가능한 단계}
+If a delegated task must create, inspect, or modify Claude compatibility artifacts under `.claude/**`, `.claude/outputs/**`, or `templates/.claude/**`, include this block in the delegated prompt:
 
----
-_🤝 Project Colleague perspective — `/professor-triage` v2.2.0_
+```text
+Sensitive-path compatibility note:
+1. Keep Codex-native .codex/** artifacts on the normal file-write path.
+2. On Claude Code v2.1.121+ with bypassPermissions, direct writes to .claude/skills/, .claude/agents/, and .claude/commands/ are allowed.
+3. On Claude Code v2.1.126+ with bypassPermissions, broader protected paths such as .claude/**, .git/**, and .vscode/** are also covered.
+4. Use /tmp/professor-triage-<timestamp>.md only as a legacy fallback when the runtime is older or still prompts, then verify the resulting diff.
 ```
 
-Note: Do NOT include a "First Impressions" (첫인상) section in the Colleague Review — this was explicitly excluded per user feedback.
-
-**4C: 🎓 Professor Synthesis** — Delegate to arch-documenter (model: opus) to post GitHub comment. This phase requires 4A and 4B results as input:
-
-```
-## 🎓 Professor Synthesis
-
-### 코드베이스 검증
-| 주장 (Architect/Colleague) | 검증 | 근거 |
-|---------------------------|------|------|
-| {주장} | ✅/⚠️/❌ | {file:line 또는 git 근거} |
-
-### 합의 및 이견
-| 주제 | Architect | Colleague | 판정 |
-|------|-----------|-----------|------|
-| {주제} | {입장} | {입장} | {종합 판단} |
-
-### 우선순위 매트릭스
-| 차원 | 평가 |
-|------|------|
-| 긴급성 | {High/Medium/Low} |
-| 중요성 | {High/Medium/Low} |
-| 규모 | {XS/S/M/L/XL} |
-| 권장 순서 | {배치 내 N/M} |
-
-### 누락된 관점
-{Architect나 Colleague 모두 제기하지 않은 고려사항}
-
-### 실행 로드맵
-| 단계 | 작업 | 파일 | 의존성 |
-|------|------|------|--------|
-| 1 | {작업} | {파일} | — |
-| 2 | {작업} | {파일} | 단계 1 |
-
-### 최종 결론
-{확정적 권장 사항이 포함된 2-3문장 종합}
-
----
-_🎓 Professor Synthesis — `/professor-triage` v2.2.0_
-```
-
-**4D: Issue Triage Comment (MANDATORY)** — Every analyzed issue MUST receive a triage comment. This is not optional — even for issues created in the same session or with existing analysis. Skipping comments breaks the triage audit trail. Delegate to mgr-gitnerd to post on each analyzed issue:
-
-```
-## 🔬 Professor Triage — Codebase Analysis Result
-
-**결정**: {Close (Already Resolved) | Close (Not Applicable) | Close (Duplicate of #NNN) | Open — action required | Open — monitoring}
-**근거**: {코드베이스 분석 기반 1-2줄 요약}
-**영향 파일**: {N}개 분석 — {N}✅ {N}⚠️ {N}❌
-**리스크**: {P1/P2/P3} | **규모**: {XS/S/M/L/XL}
-**전체 리포트**: {artifact path}
-
----
-_`/professor-triage` v2.2.0에 의해 현재 코드베이스 대비 분석됨 — 관련 이슈 {N}개_
-```
-
-**4E: Artifact Report** — Delegate to arch-documenter to write:
-
-Path: `.codex/outputs/sessions/YYYY-MM-DD/professor-triage-HHmmss.md`
-
-Timestamps use local machine time (consistent with other artifact skills).
-
-Sensitive-path artifact protocol (mandatory): if a delegated agent must create or inspect Claude compatibility artifacts under `.claude/**`, `.claude/outputs/**`, or `templates/.claude/**`, include this protocol in the delegated prompt. The agent must produce artifact bodies in `/tmp/professor-triage-{timestamp}.md` first and must not call Read, Bash, Write, or Edit directly on `.claude/**` paths in unattended flows. Codex-native `.codex/outputs/**` artifacts continue to use managed file-write APIs that create parents.
-
-Template:
-```
-# Professor Triage 리포트 — YYYY-MM-DD
-
-## 분석 대상
-| # | 제목 | 라벨 | 생성일 |
-|---|------|------|--------|
-
-## 이슈별 분석
-### #NNN — title
-- **영향 파일**: N개 분석 — N✅ N⚠️ N❌
-- **아키텍처 영향**: ...
-- **구현 경로**: ...
-- **리스크/우선순위**: P1/P2/P3
-- **규모**: XS/S/M/L/XL
-- **이미 해결됨?**: Yes/No/Partial — 근거
-- **권장 조치**: ...
-
-## 교차 분석
-### 공통 패턴
-### 중복/병합 후보
-### 상충 발견사항 해결
-### 우선순위 매트릭스
-
-## 다관점 요약
-### Architect 주요 사항
-### Colleague 주요 사항
-### Professor Synthesis 핵심 포인트
-
-## 실행된 조치
-| 이슈 | 조치 | 상태 |
-
-## 보류 중인 조치 (확인 필요)
-```
-
-### Phase 4F: Comment Verification Gate
-
-Before proceeding to Phase 5, verify ALL analyzed issues received the full set of comments (Architect + Colleague + Professor Synthesis + Triage):
-```bash
-# For each issue NNN in the batch:
-gh issue view NNN --json comments --jq '.comments | map(select(.body | contains("Professor Triage"))) | length'
-# Must be >= 1 for every issue. If any is 0, go back and post.
-
-# Also verify multi-perspective comments:
-gh issue view NNN --json comments --jq '.comments | map(select(.body | contains("Senior Architect"))) | length'
-gh issue view NNN --json comments --jq '.comments | map(select(.body | contains("Project Colleague"))) | length'
-gh issue view NNN --json comments --jq '.comments | map(select(.body | contains("Professor Synthesis"))) | length'
-# All must be >= 1. If any is 0, the corresponding Phase 4A/4B/4C was skipped — go back and post.
-```
-
-### Phase 5: Act
-
-Delegate ALL GitHub operations to mgr-gitnerd.
-
-**Automatic (low-risk, reversible):**
-
-| Condition | Action |
-|-----------|--------|
-| Phase 2 found issue already resolved (with commit evidence) | `gh issue close --reason "completed"` + comment with resolving commit |
-| Cross-analysis concludes "Not Applicable" / "no action needed" | `gh issue close --reason "not planned"` |
-| Cross-analysis detects same-series duplicates | Keep latest, close others + `duplicate` label |
-| All analysis complete | Add `verify-done` label |
-| Priority assigned | Add `P1`/`P2`/`P3` label |
-
-**Confirmation required (high-risk):**
+This protocol must be inline in the delegate prompt; relying on this SKILL.md being present in the parent context is not enough.
 
-Present to user and wait for approval before executing:
+## Phase 5 Action Policy
 
-| Condition | Action | Reason |
-|-----------|--------|--------|
-| Reopen a closed issue | Propose reopen | Unintended notifications |
-| New issue creation needed | Present draft title/body | Noise prevention |
-| Epic/milestone linking | Propose link | Project structure change |
-| Issue body modification | Present edit draft | Respect original author intent |
+Automatic, low-risk actions:
+- Issue already resolved by commit: close as completed with evidence comment.
+- Not applicable to the Codex port: close as not planned with evidence.
+- Duplicate series: close older duplicate and keep the canonical issue open.
+- Completed analysis: add `verify-done` only after evidence is current.
+- Priority assignment: add `P1`/`P2`/`P3` when supported by analysis.
 
-**Ensure `verify-done` label exists**: If not, create with `gh label create "verify-done" --color "0E8A16"`.
+Confirmation required: issue reopen, new issue creation, epic rewrites, or issue body modification.
 
-## Notes
+## Permission Mode
 
-- Phase 1: Orchestrator fetches issues directly (no agent needed)
-- Phase 2: Explore agents with `model: haiku` for codebase search; orchestrator synthesizes findings
-- Phase 3: Orchestrator directly (read-only, R010 exception); opus agent for >15 issues
-- Phase 4A/4B: `arch-documenter` (sonnet) for Architect/Colleague analysis comments (parallel)
-- Phase 4C: `arch-documenter` (opus) for Professor Synthesis comment (requires 4A+4B)
-- Phase 4D: `mgr-gitnerd` for triage comment; Phase 4E: `arch-documenter` for artifact report (parallel)
-- Phase 4F: Verification gate for all 4 comment types
-- Phase 5: `mgr-gitnerd` for all GitHub operations
-- No external dependencies (omc_issue_analyzer removed in v2.0.0, multi-perspective analysis restored in v2.1.0)
+When spawning agents, explicitly pass `mode: "bypassPermissions"` if the runtime supports it. Defaults may override agent frontmatter and reintroduce permission prompts during unattended execution.

package/templates/.claude/skills/qa-lead-routing/SKILL.md

@@ -94,7 +94,7 @@ Delegate to mgr-creator with context:
 
 ## Sensitive-Path Delegation
 
-Sensitive-path artifact protocol (mandatory): if this skill delegates work that touches `.claude/**`, `.claude/outputs/**`, `templates/.claude/**`, or read-only measurements of those paths, include this protocol directly in the delegated prompt. The delegate must produce artifact bodies in `/tmp/{skill}-{timestamp}.md` first and must avoid direct Read, Bash, Write, or Edit targets under `.claude/**` in unattended flows.
+Sensitive-path compatibility note: if this skill delegates work that touches `.claude/**`, `.claude/outputs/**`, `templates/.claude/**`, or read-only measurements of those paths, keep `.codex/**` edits on the normal Codex path. On Claude Code v2.1.121+ with `bypassPermissions`, direct writes to `.claude/skills/`, `.claude/agents/`, and `.claude/commands/` are allowed; on v2.1.126+ that extends to broader protected paths. Only use `/tmp/{skill}-{timestamp}.md` as a legacy fallback when the target runtime is older or still prompts.
 
 ## Usage
 

package/templates/.claude/skills/research/SKILL.md

@@ -204,6 +204,11 @@ Convergence expected by round 3. Hard stop at round 30.
 2. **Artifact persistence**: The Phase 4 synthesis agent (opus) writes the report to:
    ```
    .codex/outputs/sessions/{YYYY-MM-DD}/research-{HHmmss}.md
+
+### Compatibility artifact protocol
+
+Sensitive-path compatibility note: when delegated work touches `.claude/outputs/`, `.claude/**`, or `templates/.claude/**`, keep `.codex/**` artifacts on the normal file-write path. On Claude Code v2.1.121+ with `bypassPermissions`, direct compatibility writes are allowed for `.claude/skills/`, `.claude/agents/`, and `.claude/commands/`; on v2.1.126+ broader protected paths are covered. Use `/tmp/<skill>-<timestamp>.md` only as a legacy fallback when the runtime is older or still prompts.
+
    ```
    With metadata header:
    ```markdown
@@ -214,7 +219,7 @@ Convergence expected by round 3. Hard stop at round 30.
    ---
    ```
    The agent writes the artifact file using a file-write API that creates missing parent directories; do not run any Bash directory-creation pre-step for session outputs (R010 compliance).
-   Sensitive-path artifact protocol (mandatory): if a delegated research agent must inspect or create Claude compatibility artifacts under `.claude/**`, `.claude/outputs/**`, or `templates/.claude/**`, the delegated prompt must require `/tmp/research-{timestamp}.md` as the first artifact body target and must forbid direct Read, Bash, Write, or Edit targets under `.claude/**` in unattended flows.
+   Sensitive-path compatibility note: delegated research agents should keep `.codex/**` artifacts on the normal write path and use `/tmp/research-{timestamp}.md` only as a legacy fallback when an older Claude Code runtime still prompts on compatibility paths.
 3. GitHub issue auto-created with findings
 4. Action items with effort estimates
 

package/templates/.claude/skills/result-aggregation/SKILL.md

@@ -115,6 +115,10 @@ Secretary outputs:
   - Individual details if requested
 ```
 
+### Compatibility artifact protocol
+
+Sensitive-path compatibility note: when delegated work touches `.claude/outputs/`, `.claude/**`, or `templates/.claude/**`, keep `.codex/**` artifacts on the normal file-write path. On Claude Code v2.1.121+ with `bypassPermissions`, direct compatibility writes are allowed for `.claude/skills/`, `.claude/agents/`, and `.claude/commands/`; on v2.1.126+ broader protected paths are covered. Use `/tmp/<skill>-<timestamp>.md` only as a legacy fallback when the runtime is older or still prompts.
+
 ### With Ecomode
 
 ```

package/templates/.claude/skills/roundtable-debate/SKILL.md

@@ -22,7 +22,7 @@ Run a bounded debate when convergence would hide useful disagreement. Unlike `ag
 
 ## Sensitive-Path Delegation
 
-Sensitive-path artifact protocol (mandatory): if this skill delegates work that touches `.claude/**`, `.claude/outputs/**`, `templates/.claude/**`, or read-only measurements of those paths, include this protocol directly in the delegated prompt. The delegate must produce artifact bodies in `/tmp/{skill}-{timestamp}.md` first and must avoid direct Read, Bash, Write, or Edit targets under `.claude/**` in unattended flows.
+Sensitive-path compatibility note: if this skill delegates work that touches `.claude/**`, `.claude/outputs/**`, `templates/.claude/**`, or read-only measurements of those paths, keep `.codex/**` edits on the normal Codex path. On Claude Code v2.1.121+ with `bypassPermissions`, direct writes to `.claude/skills/`, `.claude/agents/`, and `.claude/commands/` are allowed; on v2.1.126+ that extends to broader protected paths. Only use `/tmp/{skill}-{timestamp}.md` as a legacy fallback when the target runtime is older or still prompts.
 
 ## Workflow
 

package/templates/.claude/skills/scout/SKILL.md

@@ -13,7 +13,7 @@ Analyze an external URL (tech blog, tool, library, methodology) to evaluate its
 
 ## Sensitive-Path Delegation
 
-Sensitive-path artifact protocol (mandatory): if this skill delegates work that touches `.claude/**`, `.claude/outputs/**`, `templates/.claude/**`, or read-only measurements of those paths, include this protocol directly in the delegated prompt. The delegate must produce artifact bodies in `/tmp/{skill}-{timestamp}.md` first and must avoid direct Read, Bash, Write, or Edit targets under `.claude/**` in unattended flows.
+Sensitive-path compatibility note: if this skill delegates work that touches `.claude/**`, `.claude/outputs/**`, `templates/.claude/**`, or read-only measurements of those paths, keep `.codex/**` edits on the normal Codex path. On Claude Code v2.1.121+ with `bypassPermissions`, direct writes to `.claude/skills/`, `.claude/agents/`, and `.claude/commands/` are allowed; on v2.1.126+ that extends to broader protected paths. Only use `/tmp/{skill}-{timestamp}.md` as a legacy fallback when the target runtime is older or still prompts.
 
 ## Usage
 

package/templates/.claude/skills/secretary-routing/SKILL.md

@@ -134,7 +134,7 @@ Evaluate: Is this a specialized management/tooling task?
 
 ## Sensitive-Path Delegation
 
-Sensitive-path artifact protocol (mandatory): if this skill delegates work that touches `.claude/**`, `.claude/outputs/**`, `templates/.claude/**`, or read-only measurements of those paths, include this protocol directly in the delegated prompt. The delegate must produce artifact bodies in `/tmp/{skill}-{timestamp}.md` first and must avoid direct Read, Bash, Write, or Edit targets under `.claude/**` in unattended flows.
+Sensitive-path compatibility note: if this skill delegates work that touches `.claude/**`, `.claude/outputs/**`, `templates/.claude/**`, or read-only measurements of those paths, keep `.codex/**` edits on the normal Codex path. On Claude Code v2.1.121+ with `bypassPermissions`, direct writes to `.claude/skills/`, `.claude/agents/`, and `.claude/commands/` are allowed; on v2.1.126+ that extends to broader protected paths. Only use `/tmp/{skill}-{timestamp}.md` as a legacy fallback when the target runtime is older or still prompts.
 
 ## Usage
 

package/templates/.claude/skills/skill-extractor/SKILL.md

@@ -146,6 +146,10 @@ The `skill-extractor-analyzer.sh` Stop hook provides a lightweight pre-analysis:
 - Emits advisory stderr message if candidates found
 - Does NOT create skills (that requires user approval via the skill)
 
+## Compatibility Artifact Protocol
+
+Sensitive-path compatibility note: when delegated work touches `.claude/outputs/`, `.claude/**`, or `templates/.claude/**`, keep `.codex/**` artifacts on the normal file-write path. On Claude Code v2.1.121+ with `bypassPermissions`, direct compatibility writes are allowed for `.claude/skills/`, `.claude/agents/`, and `.claude/commands/`; on v2.1.126+ broader protected paths are covered. Use `/tmp/<skill>-<timestamp>.md` only as a legacy fallback when the runtime is older or still prompts.
+
 ## Safety
 
 - **User approval required**: Never auto-creates skills

package/templates/.claude/skills/task-decomposition/SKILL.md

@@ -12,7 +12,7 @@ Analyzes task complexity and decomposes large tasks into smaller, parallelizable
 
 ## Sensitive-Path Delegation
 
-Sensitive-path artifact protocol (mandatory): if this skill delegates work that touches `.claude/**`, `.claude/outputs/**`, `templates/.claude/**`, or read-only measurements of those paths, include this protocol directly in the delegated prompt. The delegate must produce artifact bodies in `/tmp/{skill}-{timestamp}.md` first and must avoid direct Read, Bash, Write, or Edit targets under `.claude/**` in unattended flows.
+Sensitive-path compatibility note: if this skill delegates work that touches `.claude/**`, `.claude/outputs/**`, `templates/.claude/**`, or read-only measurements of those paths, keep `.codex/**` edits on the normal Codex path. On Claude Code v2.1.121+ with `bypassPermissions`, direct writes to `.claude/skills/`, `.claude/agents/`, and `.claude/commands/` are allowed; on v2.1.126+ that extends to broader protected paths. Only use `/tmp/{skill}-{timestamp}.md` as a legacy fallback when the target runtime is older or still prompts.
 
 ## Trigger Conditions
 

package/templates/.claude/skills/worker-reviewer-pipeline/SKILL.md

@@ -14,7 +14,7 @@ Defines an iterative Worker→Reviewer pipeline where one agent implements chang
 
 ## Sensitive-Path Delegation
 
-Sensitive-path artifact protocol (mandatory): if this skill delegates work that touches `.claude/**`, `.claude/outputs/**`, `templates/.claude/**`, or read-only measurements of those paths, include this protocol directly in the delegated prompt. The delegate must produce artifact bodies in `/tmp/{skill}-{timestamp}.md` first and must avoid direct Read, Bash, Write, or Edit targets under `.claude/**` in unattended flows.
+Sensitive-path compatibility note: if this skill delegates work that touches `.claude/**`, `.claude/outputs/**`, `templates/.claude/**`, or read-only measurements of those paths, keep `.codex/**` edits on the normal Codex path. On Claude Code v2.1.121+ with `bypassPermissions`, direct writes to `.claude/skills/`, `.claude/agents/`, and `.claude/commands/` are allowed; on v2.1.126+ that extends to broader protected paths. Only use `/tmp/{skill}-{timestamp}.md` as a legacy fallback when the target runtime is older or still prompts.
 
 ## When to Activate
 

package/templates/manifest.json

@@ -1,5 +1,10 @@
 {
-  "version": "0.4.11",
+  "version": "0.4.13",
+  "requiresCC": ">=2.1.121",
+  "claudeCode": {
+    "minimumVersion": "2.1.121",
+    "protectedPathBypassVersion": "2.1.126"
+  },
   "lastUpdated": "2026-04-28T00:01:33.302Z",
   "components": [
     {

package/templates/workflows/auto-dev.yaml

@@ -40,9 +40,9 @@ steps:
 
       Codex-native sensitive-path policy:
       - Codex-managed `.codex/` edits use the normal edit/patch flow.
-      - Do not adopt upstream Claude-only `/tmp` bypass guidance as the default path.
+      - Claude Code `bypassPermissions` can write `.claude/skills/`, `.claude/agents/`, and `.claude/commands/` directly on v2.1.121+, with broader protected-path writes available on v2.1.126+.
       - If a port requires `.claude` template parity changes, make the change explicit and verify sensitive-path guard tests.
-      - Sensitive-path artifact protocol (mandatory): delegated prompts that touch `.claude/**`, `templates/.claude/**`, `.claude/outputs/**`, or read-only measurements of those paths must instruct the subagent to produce artifact bodies in `/tmp/{skill}-{timestamp}.md` first and avoid direct Read/Bash/Write/Edit targets under `.claude/**` in unattended flows.
+      - Use `/tmp/{skill}-{timestamp}.md` only as a historical fallback for older Claude Code versions, non-bypass sessions, or runs that still surface a protected-path prompt.
     description: Execute implementation plan with appropriate agents
     foreach: planned-issue