oh-my-customcode 0.37.2 → 0.38.0

package/templates/.claude/skills/sauron-watch/SKILL.md

@@ -53,6 +53,59 @@ Ensure complete synchronization of agents, skills, documentation, and project st
 □ All routing skills have complete agent mappings
 ```
 
+#### Structural Lint Rules
+
+In addition to workflow alignment checks, verify these structural invariants:
+
+**Lint 1: Routing Coverage**
+```
+For each agent in .claude/agents/*.md:
+  Check if agent name appears in at least one routing skill:
+    - secretary-routing/SKILL.md
+    - dev-lead-routing/SKILL.md
+    - de-lead-routing/SKILL.md
+    - qa-lead-routing/SKILL.md
+
+  If not found in any routing skill:
+    WARN: "{agent} is unreachable — not referenced in any routing skill"
+```
+
+**Lint 2: Orphan Skill Detection**
+```
+For each skill in .claude/skills/*/SKILL.md:
+  Check if skill name appears in:
+    - Any agent's skills: [] frontmatter field
+    - Any routing skill's content
+    - CLAUDE.md command table (for user-invocable skills)
+
+  If not referenced anywhere:
+    WARN: "{skill} is orphaned — not referenced by any agent or routing skill"
+```
+
+**Lint 3: Circular Dependency Check**
+```
+Build dependency graph:
+  For each agent → extract skills: [] references
+  For each skill → extract agent references in body
+
+  Detect cycles: agent-A → skill-X → agent-A
+  If cycle found:
+    ERROR: "Circular dependency: {cycle path}"
+```
+
+**Lint 4: Context Fork Cap Verification**
+```
+Count skills with context: fork in frontmatter:
+  grep "context: fork" .claude/skills/*/SKILL.md
+
+  If count > 10:
+    ERROR: "Context fork cap exceeded: {count}/10"
+  If count >= 8:
+    WARN: "Context fork usage high: {count}/10 — only {10-count} slots remaining"
+```
+
+All structural lints are **advisory** (WARN level) except circular dependencies and fork cap exceeded (ERROR level — should block commit).
+
 #### Deep Round 2: Reference Verification
 ```
 □ All skill references exist
@@ -73,6 +126,34 @@ Ensure complete synchronization of agents, skills, documentation, and project st
 □ All MUST rules enforced, SHOULD rules recommended
 ```
 
+#### Spec Density Check (Advisory)
+
+Check each agent's body-to-skill ratio to detect agents with too much inline implementation detail:
+
+```
+For each agent in .claude/agents/*.md:
+  agent_body_LOC = count non-frontmatter, non-blank lines in agent body
+  referenced_skills = extract skills from frontmatter skills: field
+  total_skill_LOC = sum of LOC in each referenced SKILL.md
+
+  if total_skill_LOC > 0:
+    spec_density = agent_body_LOC / total_skill_LOC
+    if spec_density > 0.5:
+      WARN: "{agent} has spec density {spec_density:.2f} — consider extracting inline details to skills"
+  else:
+    if agent_body_LOC > 50:
+      INFO: "{agent} has {agent_body_LOC} LOC with no skill references — consider creating skills"
+```
+
+**Thresholds**:
+| Density | Status | Meaning |
+|---------|--------|---------|
+| ≤ 0.3 | Good | Agent properly delegates to skills |
+| 0.3-0.5 | OK | Acceptable inline detail |
+| > 0.5 | WARNING | Too much implementation in agent body |
+
+This check is **advisory only** — it does not block commits.
+
 ### Phase 2.5: Documentation Accuracy
 ```
 □ Every agent name in CLAUDE.md matches actual filename

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

@@ -56,26 +56,11 @@ batch    → multiple (parallel)
 
 ### Ontology-RAG Enrichment (R019)
 
-After agent selection, enrich the spawned agent's prompt with ontology context:
+If `get_agent_for_task` MCP tool is available, call it with the original query and inject `suggested_skills` into the agent prompt. Skip silently on failure.
 
-1. Call `get_agent_for_task(original_query)` via MCP
-2. Extract `suggested_skills` from response
-3. If `suggested_skills` non-empty, prepend to spawned agent prompt:
-   `"Ontology context suggests these skills may be relevant: {suggested_skills}"`
-4. On MCP failure: skip silently, proceed with unmodified prompt
+### Step 5: Soul Injection (R006)
 
-**This step is advisory only — it never changes which agent is selected.**
-
-### Step 5: Soul Injection
-
-If the selected agent has `soul: true` in its frontmatter:
-
-1. Read `.claude/agents/souls/{agent-name}.soul.md`
-2. If file exists, prepend soul content to the agent's prompt:
-   `"Identity context:\n{soul content}\n\n---\n\n"`
-3. If file doesn't exist → skip silently (no error, no injection)
-
-**This step runs after ontology-RAG enrichment. Soul content is identity context, not capability instructions.**
+If the selected agent has `soul: true` in frontmatter, read and prepend `.claude/agents/souls/{agent-name}.soul.md` content to the prompt. Skip silently if file doesn't exist.
 
 ## Routing Rules
 

package/templates/.claude/skills/structured-dev-cycle/SKILL.md

@@ -25,6 +25,21 @@ Inspired by Pi Coding Agent Workflow Extension's structured development approach
 | 5 | **Compound** | Read, Bash (tests only) | Write, Edit | Integration testing, cross-module validation |
 | 6 | **Done** | Read | Write, Edit, Bash | Summary and documentation |
 
+### Stage Model Recommendations
+
+Following the [reasoning-sandwich](/skills/reasoning-sandwich) pattern:
+
+| Stage | Recommended Model | Rationale |
+|-------|------------------|-----------|
+| 1: Plan | opus | Architectural reasoning, requirement analysis |
+| 2: Verify Plan | opus | Edge case detection, alternative evaluation |
+| 3: Implement | sonnet | Code generation, file creation optimized |
+| 4: Verify Implementation | sonnet | Test execution, structural review |
+| 5: Compound | sonnet | Integration testing, cross-module validation |
+| 6: Done | haiku | Checklist validation, summary generation |
+
+Model selection is advisory — the orchestrator may override based on task complexity.
+
 ## Stage Tracking
 
 Stage state is tracked via a marker file for hook enforcement:
@@ -80,9 +95,11 @@ A PreToolUse hook in `.claude/hooks/hooks.json` checks this marker and blocks Wr
 
 **Codex-Exec Hybrid Option**: When entering Stage 3:
 1. Check `/tmp/.claude-env-status-*` for codex CLI availability
-2. If available AND task involves new file creation:
-   - Suggest: `[Codex Hybrid Available] New file generation can use codex-exec for faster scaffolding. Orchestrator may delegate initial code generation to codex-exec, then have Claude expert review.`
-3. If unavailable → proceed with standard implementation via Claude experts
+2. If available AND task involves new file creation → automatically delegate scaffolding to `/codex-exec`:
+   - Display: `[Codex Hybrid] Delegating scaffolding to codex-exec...`
+   - codex-exec generates initial code (strength: fast generation)
+   - Claude expert reviews and refines codex output (strength: reasoning, quality)
+3. If unavailable → display `[Codex] Unavailable — proceeding with Claude experts directly` and proceed with standard implementation via Claude experts
 
 Suitable for codex hybrid: new files, boilerplate, test stubs, scaffolding
 Not suitable: modifying existing code, architecture-dependent changes

package/templates/guides/claude-code/index.yaml

@@ -49,3 +49,8 @@ documents:
     description: 프로덕션 에이전트 운영
     path: ./10-monitoring.md
     status: placeholder  # 공식 문서 미공개
+
+  - name: workflow-patterns
+    title: 워크플로우 패턴
+    description: AI 에이전트 작업에 적합한 워크플로우 패턴 선택 가이드
+    path: ./12-workflow-patterns.md

package/templates/manifest.json

@@ -1,12 +1,12 @@
 {
-  "version": "0.37.2",
-  "lastUpdated": "2026-03-14T00:00:00.000Z",
+  "version": "0.38.0",
+  "lastUpdated": "2026-03-16T00:00:00.000Z",
   "components": [
     {
       "name": "rules",
       "path": ".claude/rules",
       "description": "Agent behavior rules and guidelines",
-      "files": 19
+      "files": 20
     },
     {
       "name": "agents",

package/templates/.claude/hooks/scripts/session-compliance-report.sh

@@ -1,65 +0,0 @@
-#!/bin/bash
-# Stop hook: Session compliance report (R265 Phase 1)
-# Reads violation logs collected by PreToolUse hooks during the session
-# Advisory only — never blocks session termination
-# Ref: https://github.com/baekenough/oh-my-customcode/issues/265
-
-set -euo pipefail
-
-input=$(cat)
-
-VIOLATIONS_FILE="/tmp/.claude-violations-${PPID}"
-TASK_COUNT_FILE="/tmp/.claude-task-count-${PPID}"
-
-echo "" >&2
-echo "╔══════════════════════════════════════════════╗" >&2
-echo "║        Session Compliance Report             ║" >&2
-echo "╚══════════════════════════════════════════════╝" >&2
-
-# Count total Agent/Task calls
-if [ -f "$TASK_COUNT_FILE" ]; then
-  TOTAL_TASKS=$(cat "$TASK_COUNT_FILE")
-  echo "[Compliance] Agent/Task calls this session: ${TOTAL_TASKS}" >&2
-else
-  TOTAL_TASKS=0
-  echo "[Compliance] Agent/Task calls this session: 0" >&2
-fi
-
-# Check violations
-if [ -f "$VIOLATIONS_FILE" ] && [ -s "$VIOLATIONS_FILE" ]; then
-  VIOLATION_COUNT=$(wc -l < "$VIOLATIONS_FILE" | tr -d ' ')
-  echo "[Compliance] Violations detected: ${VIOLATION_COUNT}" >&2
-  echo "" >&2
-
-  # Group by rule
-  R010_COUNT=$(grep -c '"rule":"R010"' "$VIOLATIONS_FILE" 2>/dev/null || echo "0")
-  R018_COUNT=$(grep -c '"rule":"R018"' "$VIOLATIONS_FILE" 2>/dev/null || echo "0")
-
-  if [ "$R010_COUNT" -gt 0 ]; then
-    echo "  R010 (Git Delegation): ${R010_COUNT} violation(s)" >&2
-    grep '"rule":"R010"' "$VIOLATIONS_FILE" | jq -r '.detail' 2>/dev/null | while read -r detail; do
-      echo "    - ${detail}" >&2
-    done
-  fi
-
-  if [ "$R018_COUNT" -gt 0 ]; then
-    echo "  R018 (Agent Teams): ${R018_COUNT} violation(s)" >&2
-    grep '"rule":"R018"' "$VIOLATIONS_FILE" | jq -r '.detail' 2>/dev/null | while read -r detail; do
-      echo "    - ${detail}" >&2
-    done
-  fi
-
-  echo "" >&2
-  echo "[Compliance] Review violations above and consider rule updates per R016." >&2
-else
-  echo "[Compliance] No violations detected. All clear!" >&2
-fi
-
-echo "────────────────────────────────────────────────" >&2
-
-# Cleanup temp files (best effort)
-rm -f "$VIOLATIONS_FILE" 2>/dev/null || true
-
-# CRITICAL: Always pass through input and exit 0
-echo "$input"
-exit 0