@su-record/vibe 2.4.34 → 2.4.36

package/skills/tool-fallback.md

@@ -1,126 +1,126 @@
----
-description: Tool failure fallback strategies with circuit breaker. Auto-activates on API errors, search failures, timeouts, 429, 5xx, overloaded errors.
----
-# Tool Fallback Strategies
-
-Guide for finding alternatives when tools fail to continue work.
-
-## Circuit Breaker Pattern
-
-Track tool failures and temporarily disable unreliable tools:
-
-```
-Tool State Machine:
-┌─────────┐   3 failures   ┌─────────┐   30s cooldown   ┌─────────────┐
-│ CLOSED  │ ─────────────→ │  OPEN   │ ───────────────→ │ HALF-OPEN   │
-│ (normal)│                │ (block) │                  │ (test 1 req)│
-└─────────┘                └─────────┘                  └─────────────┘
-     ↑                                                        │
-     │                      success                           │
-     └────────────────────────────────────────────────────────┘
-```
-
-| State | Behavior |
-|-------|----------|
-| **CLOSED** | Normal operation, count failures |
-| **OPEN** | Block all requests, use alternative immediately |
-| **HALF-OPEN** | Allow 1 test request after cooldown |
-
-**Per-tool tracking:**
-```
-GPT hook: failures=2, state=CLOSED
-Gemini hook: failures=0, state=CLOSED
-WebSearch: failures=3, state=OPEN (blocked until 14:32:00)
-context7: failures=1, state=CLOSED
-```
-
-## When Web Search Fails
-
-| Alternative | Method |
-|-------------|--------|
-| context7 MCP | `mcp__context7__query-docs` |
-| GPT hook | `gpt- [your question]` |
-| Cached knowledge | Use Claude's built-in knowledge |
-
-```
-Web Search fails (429, 529, timeout)
-    ↓
-Check circuit breaker state
-    ↓
-If OPEN → Skip to alternative immediately
-    ↓
-If CLOSED → Try context7 for library docs
-    ↓
-If still fails: gpt- [question]
-    ↓
-Last resort: Claude's built-in knowledge
-```
-
-## API Error Responses
-
-| Error | Cause | Response | Circuit Breaker |
-|-------|-------|----------|-----------------|
-| 429 | Rate Limit | Exponential backoff | +1 failure count |
-| 5xx | Server Error | Switch to alternative | +1 failure count |
-| 529 | Overloaded | Wait and retry | +1 failure count |
-| Timeout | Network | Split request or retry | +1 failure count |
-| 401/403 | Auth Error | Re-auth or alternative | Don't count (auth issue) |
-
-## When File/Code Not Found
-
-```
-Glob fails
-    ↓
-Expand pattern: *.ts → **/*.ts → **/*
-    ↓
-Use Grep for content-based search
-    ↓
-Check git log for history
-```
-
-## When LLM Hook Fails
-
-```
-gpt- or gpt. [question] fails
-    ↓
-Check circuit: If OPEN, skip to next
-    ↓
-Try gemini- or gemini. [question] (similar capability)
-    ↓
-Try context7 (for docs)
-    ↓
-Claude solves alone
-```
-
-## Retry Strategy with Circuit Breaker
-
-```
-Request to tool
-    ↓
-Check circuit state
-    ↓
-┌─ OPEN? ──→ Use alternative immediately (no retry)
-│
-└─ CLOSED/HALF-OPEN? ──→ Try request
-                              ↓
-                         Success? ──→ Reset failure count
-                              ↓ No
-                         Retry with backoff:
-                           retry(1): wait 2s
-                           retry(2): wait 4s
-                           retry(3): wait 8s
-                              ↓
-                         All failed? ──→ +1 failure, check threshold
-                              ↓
-                         failures >= 3? ──→ OPEN circuit for 30s
-                              ↓
-                         Use alternative
-```
-
-## Principles
-
-1. **Never stop** - Always find an alternative
-2. **Before asking user** - Try alternatives first
-3. **Track failures** - Open circuit after 3 consecutive failures
-4. **Auto-recover** - Test after 30s cooldown
-5. **Fail fast** - Skip blocked tools immediately
+---
+description: Tool failure fallback strategies with circuit breaker. Auto-activates on API errors, search failures, timeouts, 429, 5xx, overloaded errors.
+---
+# Tool Fallback Strategies
+
+Guide for finding alternatives when tools fail to continue work.
+
+## Circuit Breaker Pattern
+
+Track tool failures and temporarily disable unreliable tools:
+
+```
+Tool State Machine:
+┌─────────┐   3 failures   ┌─────────┐   30s cooldown   ┌─────────────┐
+│ CLOSED  │ ─────────────→ │  OPEN   │ ───────────────→ │ HALF-OPEN   │
+│ (normal)│                │ (block) │                  │ (test 1 req)│
+└─────────┘                └─────────┘                  └─────────────┘
+     ↑                                                        │
+     │                      success                           │
+     └────────────────────────────────────────────────────────┘
+```
+
+| State | Behavior |
+|-------|----------|
+| **CLOSED** | Normal operation, count failures |
+| **OPEN** | Block all requests, use alternative immediately |
+| **HALF-OPEN** | Allow 1 test request after cooldown |
+
+**Per-tool tracking:**
+```
+GPT hook: failures=2, state=CLOSED
+Gemini hook: failures=0, state=CLOSED
+WebSearch: failures=3, state=OPEN (blocked until 14:32:00)
+context7: failures=1, state=CLOSED
+```
+
+## When Web Search Fails
+
+| Alternative | Method |
+|-------------|--------|
+| context7 MCP | `mcp__context7__query-docs` |
+| GPT hook | `gpt- [your question]` |
+| Cached knowledge | Use Claude's built-in knowledge |
+
+```
+Web Search fails (429, 529, timeout)
+    ↓
+Check circuit breaker state
+    ↓
+If OPEN → Skip to alternative immediately
+    ↓
+If CLOSED → Try context7 for library docs
+    ↓
+If still fails: gpt- [question]
+    ↓
+Last resort: Claude's built-in knowledge
+```
+
+## API Error Responses
+
+| Error | Cause | Response | Circuit Breaker |
+|-------|-------|----------|-----------------|
+| 429 | Rate Limit | Exponential backoff | +1 failure count |
+| 5xx | Server Error | Switch to alternative | +1 failure count |
+| 529 | Overloaded | Wait and retry | +1 failure count |
+| Timeout | Network | Split request or retry | +1 failure count |
+| 401/403 | Auth Error | Re-auth or alternative | Don't count (auth issue) |
+
+## When File/Code Not Found
+
+```
+Glob fails
+    ↓
+Expand pattern: *.ts → **/*.ts → **/*
+    ↓
+Use Grep for content-based search
+    ↓
+Check git log for history
+```
+
+## When LLM Hook Fails
+
+```
+gpt- or gpt. [question] fails
+    ↓
+Check circuit: If OPEN, skip to next
+    ↓
+Try gemini- or gemini. [question] (similar capability)
+    ↓
+Try context7 (for docs)
+    ↓
+Claude solves alone
+```
+
+## Retry Strategy with Circuit Breaker
+
+```
+Request to tool
+    ↓
+Check circuit state
+    ↓
+┌─ OPEN? ──→ Use alternative immediately (no retry)
+│
+└─ CLOSED/HALF-OPEN? ──→ Try request
+                              ↓
+                         Success? ──→ Reset failure count
+                              ↓ No
+                         Retry with backoff:
+                           retry(1): wait 2s
+                           retry(2): wait 4s
+                           retry(3): wait 8s
+                              ↓
+                         All failed? ──→ +1 failure, check threshold
+                              ↓
+                         failures >= 3? ──→ OPEN circuit for 30s
+                              ↓
+                         Use alternative
+```
+
+## Principles
+
+1. **Never stop** - Always find an alternative
+2. **Before asking user** - Try alternatives first
+3. **Track failures** - Open circuit after 3 consecutive failures
+4. **Auto-recover** - Test after 30s cooldown
+5. **Fail fast** - Skip blocked tools immediately

package/skills/vibe-capabilities.md

@@ -1,129 +1,129 @@
----
-description: Vibe framework capabilities overview. Auto-activates when working on vibe projects, asking about available features, or needing workflow guidance.
----
-# Vibe Capabilities
-
-Complete guide to vibe's scenario-driven development framework.
-
-## Core Workflow
-
-```
-/vibe.spec → /vibe.run → /vibe.verify → /vibe.review
-```
-
-## Slash Commands
-
-| Command | Purpose |
-|---------|---------|
-| `/vibe.spec "feature"` | Create SPEC with PTCF structure + parallel research |
-| `/vibe.run "feature"` | Implement based on SPEC |
-| `/vibe.run "feature" ultrawork` | Maximum performance mode |
-| `/vibe.verify "feature"` | BDD scenario verification |
-| `/vibe.review` | Parallel code review (13+ agents) |
-| `/vibe.analyze` | Project analysis |
-| `/vibe.reason "problem"` | Systematic reasoning |
-| `/vibe.utils --e2e` | Playwright E2E testing |
-| `/vibe.utils --compound` | Document solutions |
-
-## Built-in Tools
-
-### Semantic Code Analysis
-
-```bash
-node -e "import('@su-record/vibe/tools').then(t => t.TOOL_NAME({...}))"
-```
-
-| Tool | Purpose |
-|------|---------|
-| `findSymbol` | Find symbol definitions |
-| `findReferences` | Find all references |
-| `analyzeComplexity` | Code complexity analysis |
-| `validateCodeQuality` | Quality validation |
-
-### Memory Management
-
-| Tool | Purpose |
-|------|---------|
-| `startSession` | Start session (restore previous context) |
-| `autoSaveContext` | Save current state |
-| `saveMemory` | Save important decisions |
-| `recallMemory` | Recall saved memory |
-| `listMemories` | List saved memories |
-
-## Orchestrator
-
-### Parallel Research (4 agents)
-
-```bash
-node -e "import('@su-record/vibe/orchestrator').then(o =>
-  o.research('feature', ['stack1', 'stack2'])
-)"
-```
-
-### Background Agents
-
-```bash
-node -e "import('@su-record/vibe/orchestrator').then(o =>
-  o.launchBackgroundAgent({ prompt: '...', agentName: '...' })
-)"
-```
-
-## LLM Integrations (Hooks)
-
-Use prefix patterns to call GPT/Gemini:
-
-| LLM | Prefix | Example | Features |
-|-----|--------|---------|----------|
-| GPT | `gpt-`, `gpt.`, `지피티-` | `gpt.오늘 서울 날씨` | Web Search enabled |
-| Gemini | `gemini-`, `gemini.`, `제미나이-` | `gemini.Analyze UX` | Google Search enabled |
-| context7 | MCP call | Library documentation | - |
-
-## ULTRAWORK Mode
-
-Add `ultrawork` or `ulw` for maximum performance:
-
-- Parallel subagent exploration (3+ concurrent)
-- Background agents preparing next phase
-- Phase pipelining (no wait between phases)
-- Boulder Loop (auto-progress until complete)
-- Auto-retry on errors (max 3 times)
-- Auto-compress at 70%+ context
-
-## Review Agents (13+)
-
-| Category | Agents |
-|----------|--------|
-| Security | security-reviewer, data-integrity-reviewer |
-| Performance | performance-reviewer, complexity-reviewer |
-| Architecture | architecture-reviewer, simplicity-reviewer |
-| Language | python, typescript, rails, react reviewers |
-| Context | git-history, test-coverage reviewers |
-
-## Project Structure
-
-```
-.claude/
-├── commands/          # Slash commands
-├── agents/            # Sub-agents
-├── skills/            # Auto-activated guides
-└── vibe/
-    ├── specs/         # SPEC documents
-    ├── features/      # BDD scenarios
-    ├── rules/         # Coding rules
-    └── solutions/     # Documented solutions
-```
-
-## Context Management
-
-- Session start: Auto-restore previous context
-- 70/80/90%: Auto-save checkpoints
-- Context overflow: Use `saveMemory` → `/new`
-
-## Without Commands
-
-Even without `/vibe.*` commands, you can:
-
-1. Call tools directly via node commands
-2. Use LLM hooks (`gpt-`, `gemini-`)
-3. Reference skills for guidance
-4. Apply coding rules from `.claude/vibe/rules/`
+---
+description: Vibe framework capabilities overview. Auto-activates when working on vibe projects, asking about available features, or needing workflow guidance.
+---
+# Vibe Capabilities
+
+Complete guide to vibe's scenario-driven development framework.
+
+## Core Workflow
+
+```
+/vibe.spec → /vibe.run → /vibe.verify → /vibe.review
+```
+
+## Slash Commands
+
+| Command | Purpose |
+|---------|---------|
+| `/vibe.spec "feature"` | Create SPEC with PTCF structure + parallel research |
+| `/vibe.run "feature"` | Implement based on SPEC |
+| `/vibe.run "feature" ultrawork` | Maximum performance mode |
+| `/vibe.verify "feature"` | BDD scenario verification |
+| `/vibe.review` | Parallel code review (13+ agents) |
+| `/vibe.analyze` | Project analysis |
+| `/vibe.reason "problem"` | Systematic reasoning |
+| `/vibe.utils --e2e` | Playwright E2E testing |
+| `/vibe.utils --compound` | Document solutions |
+
+## Built-in Tools
+
+### Semantic Code Analysis
+
+```bash
+node -e "import('@su-record/vibe/tools').then(t => t.TOOL_NAME({...}))"
+```
+
+| Tool | Purpose |
+|------|---------|
+| `findSymbol` | Find symbol definitions |
+| `findReferences` | Find all references |
+| `analyzeComplexity` | Code complexity analysis |
+| `validateCodeQuality` | Quality validation |
+
+### Memory Management
+
+| Tool | Purpose |
+|------|---------|
+| `startSession` | Start session (restore previous context) |
+| `autoSaveContext` | Save current state |
+| `saveMemory` | Save important decisions |
+| `recallMemory` | Recall saved memory |
+| `listMemories` | List saved memories |
+
+## Orchestrator
+
+### Parallel Research (4 agents)
+
+```bash
+node -e "import('@su-record/vibe/orchestrator').then(o =>
+  o.research('feature', ['stack1', 'stack2'])
+)"
+```
+
+### Background Agents
+
+```bash
+node -e "import('@su-record/vibe/orchestrator').then(o =>
+  o.launchBackgroundAgent({ prompt: '...', agentName: '...' })
+)"
+```
+
+## LLM Integrations (Hooks)
+
+Use prefix patterns to call GPT/Gemini:
+
+| LLM | Prefix | Example | Features |
+|-----|--------|---------|----------|
+| GPT | `gpt-`, `gpt.`, `지피티-` | `gpt.오늘 서울 날씨` | Web Search enabled |
+| Gemini | `gemini-`, `gemini.`, `제미나이-` | `gemini.Analyze UX` | Google Search enabled |
+| context7 | MCP call | Library documentation | - |
+
+## ULTRAWORK Mode
+
+Add `ultrawork` or `ulw` for maximum performance:
+
+- Parallel subagent exploration (3+ concurrent)
+- Background agents preparing next phase
+- Phase pipelining (no wait between phases)
+- Boulder Loop (auto-progress until complete)
+- Auto-retry on errors (max 3 times)
+- Auto-compress at 70%+ context
+
+## Review Agents (13+)
+
+| Category | Agents |
+|----------|--------|
+| Security | security-reviewer, data-integrity-reviewer |
+| Performance | performance-reviewer, complexity-reviewer |
+| Architecture | architecture-reviewer, simplicity-reviewer |
+| Language | python, typescript, rails, react reviewers |
+| Context | git-history, test-coverage reviewers |
+
+## Project Structure
+
+```
+.claude/
+├── commands/          # Slash commands
+├── agents/            # Sub-agents
+├── skills/            # Auto-activated guides
+└── vibe/
+    ├── specs/         # SPEC documents
+    ├── features/      # BDD scenarios
+    ├── rules/         # Coding rules
+    └── solutions/     # Documented solutions
+```
+
+## Context Management
+
+- Session start: Auto-restore previous context
+- 70/80/90%: Auto-save checkpoints
+- Context overflow: Use `saveMemory` → `/new`
+
+## Without Commands
+
+Even without `/vibe.*` commands, you can:
+
+1. Call tools directly via node commands
+2. Use LLM hooks (`gpt-`, `gemini-`)
+3. Reference skills for guidance
+4. Apply coding rules from `.claude/vibe/rules/`

package/{.claude/vibe → vibe}/config.json

@@ -1,14 +1,14 @@
 {
   "models": {
     "gpt": {
-      "enabled": true,
+      "enabled": false,
+      "authType": "oauth",
       "role": "architecture",
       "description": "아키텍처/디버깅 (GPT 5.2)"
     },
     "gemini": {
-      "enabled": true,
+      "enabled": false,
       "authType": "oauth",
-      "email": "su@fallingo.app",
       "role": "exploration",
       "description": "Gemini 3 Flash/Pro (탐색, UI/UX)"
     }