@waoooo/claude-skills 1.0.0

package/skills/multi-model-review/SKILL.md

@@ -0,0 +1,459 @@
+---
+name: multi-model-review
+description: Generic multi-model review framework. Use when you need to review content with multiple AI models (Claude, Gemini, Codex) and aggregate results with automatic decision-making. Supports OpenSpec review, Specs review, and Acceptance review.
+---
+
+# Multi-Model Review
+
+Generic framework for reviewing content with multiple AI models and aggregating results.
+
+## Core workflow
+
+### 1. Prepare review
+
+**Required inputs**:
+- `reviewType`: "openspec" | "specs" | "acceptance"
+- `contentPath`: Path to content being reviewed
+- `models`: Array of models to use (e.g., ["gemini", "codex"])
+- `thresholds`: Decision thresholds
+
+**Optional inputs**:
+- `reviewPrompt`: Custom prompt template
+- `maxRetries`: Maximum auto-fix retries (default: 3)
+
+### 2. Call multiple models in parallel
+
+```bash
+# Call Gemini
+ask gemini "Review this OpenSpec: $(cat openspec/changes/001/proposal.md)"
+
+# Call Codex
+ask codex "Review this OpenSpec: $(cat openspec/changes/001/proposal.md)"
+
+# Wait for responses
+sleep 30
+
+# Get responses
+GEMINI_RESPONSE=$(pend gemini)
+CODEX_RESPONSE=$(pend codex)
+```
+
+### 3. Aggregate results
+
+**Parse each model's response**:
+- Extract: pass/fail, score, issues, suggestions
+- Categorize issues: CRITICAL, HIGH, MEDIUM, LOW
+
+**Calculate metrics**:
+```bash
+# Pass rate
+PASS_COUNT=0
+TOTAL_COUNT=2
+
+if echo "$GEMINI_RESPONSE" | grep -q "PASS"; then
+  PASS_COUNT=$((PASS_COUNT + 1))
+fi
+
+if echo "$CODEX_RESPONSE" | grep -q "PASS"; then
+  PASS_COUNT=$((PASS_COUNT + 1))
+fi
+
+PASS_RATE=$((PASS_COUNT * 100 / TOTAL_COUNT))
+```
+
+**Aggregate issues**:
+```bash
+# Count issues by severity
+CRITICAL_ISSUES=$(echo "$GEMINI_RESPONSE $CODEX_RESPONSE" | grep -c "CRITICAL")
+HIGH_ISSUES=$(echo "$GEMINI_RESPONSE $CODEX_RESPONSE" | grep -c "HIGH")
+MEDIUM_ISSUES=$(echo "$GEMINI_RESPONSE $CODEX_RESPONSE" | grep -c "MEDIUM")
+LOW_ISSUES=$(echo "$GEMINI_RESPONSE $CODEX_RESPONSE" | grep -c "LOW")
+```
+
+### 4. Make decision based on thresholds
+
+**Default thresholds**:
+```json
+{
+  "passRate": 75,
+  "criticalIssues": 0,
+  "mediumIssues": 2
+}
+```
+
+**Decision logic**:
+```bash
+if [ $PASS_RATE -ge 75 ] && [ $CRITICAL_ISSUES -eq 0 ] && [ $MEDIUM_ISSUES -le 2 ]; then
+  DECISION="PASS"
+elif [ $RETRY_COUNT -lt 3 ]; then
+  DECISION="FIX_AND_RETRY"
+else
+  DECISION="ASK_USER"
+fi
+```
+
+### 5. Auto-fix and retry (if needed)
+
+**If decision is FIX_AND_RETRY**:
+1. Analyze issues
+2. Modify files to fix issues
+3. Increment retry count
+4. Re-run review (go to step 2)
+
+**If decision is ASK_USER**:
+1. Show summary to user
+2. Ask: "Continue fixing or stop?"
+3. If continue: go to step 5
+4. If stop: exit
+
+### 6. Save results
+
+**Directory structure**:
+```
+exports/jobs/<job-id>/reviews/<review-type>/
+├── gemini/
+│   └── review.md
+├── codex/
+│   └── review.md
+└── summary.md
+```
+
+**Summary format**:
+```markdown
+# Review Summary
+
+## Decision
+✅ PASS / ❌ FAIL / 🔄 FIX_AND_RETRY
+
+## Metrics
+- Pass Rate: 75%
+- Critical Issues: 0
+- High Issues: 1
+- Medium Issues: 2
+- Low Issues: 5
+
+## Models
+- Gemini: PASS (Score: 8/10)
+- Codex: PASS (Score: 7/10)
+
+## Issues
+### Critical
+(none)
+
+### High
+1. Missing validation for user input
+
+### Medium
+1. No error handling for API timeout
+2. Unclear description in section 3
+
+### Low
+1. Typo in line 45
+2. ...
+
+## Next Action
+- [x] Auto-fixed and retried
+- [ ] Manual fix required
+- [ ] Approved
+```
+
+## Review types
+
+### OpenSpec Review
+
+**Content**: `openspec/changes/<id>/proposal.md`
+
+**Prompt template**:
+```
+Review this OpenSpec proposal:
+
+{content}
+
+Check for:
+1. Clear problem statement
+2. Well-defined goals
+3. Feasible solution
+4. Complete requirements
+5. Risk assessment
+
+Rate: PASS/FAIL
+Score: 0-10
+Issues: List with severity (CRITICAL/HIGH/MEDIUM/LOW)
+```
+
+### Specs Review
+
+**Content**: `exports/jobs/<job-id>/specs/*.json`
+
+**Prompt template**:
+```
+Review these development specs:
+
+{content}
+
+Check for:
+1. Complete API definitions
+2. Clear data models
+3. Validation rules
+4. Error handling
+5. Test coverage
+
+Rate: PASS/FAIL
+Score: 0-10
+Issues: List with severity
+```
+
+### Acceptance Review
+
+**Content**: `exports/jobs/<job-id>/verification.json`
+
+**Prompt template**:
+```
+Review this acceptance verification:
+
+{content}
+
+Check for:
+1. All tasks completed
+2. All tests passed
+3. Documentation updated
+4. No critical issues
+5. Ready for production
+
+Rate: PASS/FAIL
+Score: 0-10
+Issues: List with severity
+```
+
+## Usage examples
+
+### Example 1: OpenSpec Review
+
+```bash
+# Step 1: Prepare
+REVIEW_TYPE="openspec"
+CONTENT_PATH="openspec/changes/001-pricing-system/proposal.md"
+MODELS=("gemini" "codex")
+
+# Step 2: Call models
+for model in "${MODELS[@]}"; do
+  ask $model "Review this OpenSpec: $(cat $CONTENT_PATH)"
+done
+
+sleep 30
+
+# Step 3: Get responses
+GEMINI_RESPONSE=$(pend gemini)
+CODEX_RESPONSE=$(pend codex)
+
+# Step 4: Aggregate
+# ... (parse and calculate metrics)
+
+# Step 5: Make decision
+if [ $PASS_RATE -ge 75 ]; then
+  echo "✅ Review PASSED"
+else
+  echo "❌ Review FAILED"
+  # Auto-fix or ask user
+fi
+
+# Step 6: Save results
+mkdir -p exports/jobs/2026-02-11-pricing-system/reviews/openspec/gemini
+echo "$GEMINI_RESPONSE" > exports/jobs/2026-02-11-pricing-system/reviews/openspec/gemini/review.md
+
+mkdir -p exports/jobs/2026-02-11-pricing-system/reviews/openspec/codex
+echo "$CODEX_RESPONSE" > exports/jobs/2026-02-11-pricing-system/reviews/openspec/codex/review.md
+
+# Generate summary
+cat > exports/jobs/2026-02-11-pricing-system/reviews/openspec/summary.md << EOF
+# OpenSpec Review Summary
+
+## Decision
+✅ PASS
+
+## Metrics
+- Pass Rate: 100%
+- Critical Issues: 0
+- Medium Issues: 1
+
+## Models
+- Gemini: PASS (8/10)
+- Codex: PASS (7/10)
+EOF
+```
+
+### Example 2: With auto-fix retry
+
+```bash
+RETRY_COUNT=0
+MAX_RETRIES=3
+
+while [ $RETRY_COUNT -lt $MAX_RETRIES ]; do
+  # Run review
+  # ... (steps 2-4)
+  
+  if [ "$DECISION" = "PASS" ]; then
+    echo "✅ Review passed"
+    break
+  elif [ "$DECISION" = "FIX_AND_RETRY" ]; then
+    echo "🔄 Auto-fixing issues (retry $((RETRY_COUNT + 1))/$MAX_RETRIES)"
+    
+    # Analyze issues and fix
+    # ... (modify files)
+    
+    RETRY_COUNT=$((RETRY_COUNT + 1))
+  else
+    echo "❌ Review failed after $RETRY_COUNT retries"
+    
+    # Ask user
+    read -p "Continue fixing? (y/n) " CONTINUE
+    if [ "$CONTINUE" != "y" ]; then
+      break
+    fi
+  fi
+done
+```
+
+### Example 3: Create Issue if failed
+
+```bash
+if [ "$DECISION" != "PASS" ]; then
+  # Create GitHub Issue
+  gh issue create \
+    --title "Review failed: $REVIEW_TYPE" \
+    --body "$(cat <<EOF
+## Review Type
+$REVIEW_TYPE
+
+## Metrics
+- Pass Rate: $PASS_RATE%
+- Critical Issues: $CRITICAL_ISSUES
+- Medium Issues: $MEDIUM_ISSUES
+
+## Context
+- Job: $(basename $(pwd))
+- Content: $CONTENT_PATH
+
+## Action Required
+Review the issues and fix them.
+
+See full review: exports/jobs/.../reviews/$REVIEW_TYPE/summary.md
+EOF
+)" \
+    --label "review-feedback,$REVIEW_TYPE,priority:high" \
+    --assignee "@me"
+fi
+```
+
+## Integration with Product Builder
+
+### Step 5: OpenSpec Review
+
+```bash
+skill(
+  name="multi-model-review",
+  params={
+    "reviewType": "openspec",
+    "contentPath": "openspec/changes/001-pricing-system/",
+    "models": ["gemini", "codex"],
+    "thresholds": {
+      "passRate": 75,
+      "criticalIssues": 0,
+      "mediumIssues": 2
+    }
+  }
+)
+```
+
+### Step 9: Specs Review
+
+```bash
+skill(
+  name="multi-model-review",
+  params={
+    "reviewType": "specs",
+    "contentPath": "exports/jobs/2026-02-11-pricing-system/specs/",
+    "models": ["gemini", "codex"],
+    "thresholds": {
+      "passRate": 75,
+      "criticalIssues": 0,
+      "mediumIssues": 2
+    }
+  }
+)
+```
+
+### Step 20: Acceptance Review
+
+```bash
+skill(
+  name="multi-model-review",
+  params={
+    "reviewType": "acceptance",
+    "contentPath": "exports/jobs/2026-02-11-pricing-system/verification.json",
+    "models": ["gemini", "codex"],
+    "thresholds": {
+      "passRate": 75,
+      "criticalIssues": 0,
+      "mediumIssues": 0
+    }
+  }
+)
+```
+
+## Important rules
+
+- **ALWAYS** call models in parallel for speed
+- **ALWAYS** wait sufficient time for responses (30+ seconds)
+- **ALWAYS** save individual model responses
+- **ALWAYS** generate summary.md
+- **ALWAYS** create Issue if review fails after max retries
+- **NEVER** skip aggregation step
+- **NEVER** ignore critical issues
+- Use auto-fix for simple issues
+- Ask user for complex issues
+
+## Error handling
+
+### Model timeout
+
+```bash
+# Wait with timeout
+timeout 60 bash -c 'while ! pend gemini 2>/dev/null; do sleep 5; done'
+
+if [ $? -eq 124 ]; then
+  echo "⚠️ Gemini timeout, skipping"
+  # Continue with other models
+fi
+```
+
+### Model error
+
+```bash
+RESPONSE=$(pend gemini)
+
+if echo "$RESPONSE" | grep -q "ERROR"; then
+  echo "⚠️ Gemini error: $RESPONSE"
+  # Continue with other models
+fi
+```
+
+### No models available
+
+```bash
+if [ ${#MODELS[@]} -eq 0 ]; then
+  echo "❌ No models available for review"
+  exit 1
+fi
+```
+
+## Notes
+
+- Works in **current directory**
+- Uses CCB (ask/pend commands)
+- Requires CCB to be configured
+- Supports any number of models
+- Customizable thresholds
+- Automatic retry with fix
+- Creates GitHub Issues for failures
+- Compatible with Product Builder workflow

package/skills/notion-automation/SKILL.md

@@ -0,0 +1,215 @@
+---
+name: notion-automation
+description: "Automate Notion tasks via Rube MCP (Composio): pages, databases, blocks, comments, users. Always search tools first for current schemas."
+requires:
+  mcp: [rube]
+---
+
+# Notion Automation via Rube MCP
+
+Automate Notion operations through Composio's Notion toolkit via Rube MCP.
+
+## Prerequisites
+
+- Rube MCP must be connected (RUBE_SEARCH_TOOLS available)
+- Active Notion connection via `RUBE_MANAGE_CONNECTIONS` with toolkit `notion`
+- Always call `RUBE_SEARCH_TOOLS` first to get current tool schemas
+
+## Setup
+
+**Get Rube MCP**: Add `https://rube.app/mcp` as an MCP server in your client configuration. No API keys needed — just add the endpoint and it works.
+
+
+1. Verify Rube MCP is available by confirming `RUBE_SEARCH_TOOLS` responds
+2. Call `RUBE_MANAGE_CONNECTIONS` with toolkit `notion`
+3. If connection is not ACTIVE, follow the returned auth link to complete Notion OAuth
+4. Confirm connection status shows ACTIVE before running any workflows
+
+## Core Workflows
+
+### 1. Create and Manage Pages
+
+**When to use**: User wants to create, update, or archive Notion pages
+
+**Tool sequence**:
+1. `NOTION_SEARCH_NOTION_PAGE` - Find parent page or existing page [Prerequisite]
+2. `NOTION_CREATE_NOTION_PAGE` - Create a new page under a parent [Optional]
+3. `NOTION_RETRIEVE_PAGE` - Get page metadata/properties [Optional]
+4. `NOTION_UPDATE_PAGE` - Update page properties, title, icon, cover [Optional]
+5. `NOTION_ARCHIVE_NOTION_PAGE` - Soft-delete (archive) a page [Optional]
+
+**Key parameters**:
+- `query`: Search text for SEARCH_NOTION_PAGE
+- `parent_id`: Parent page or database ID
+- `page_id`: Page ID for retrieval/update/archive
+- `properties`: Page property values matching parent schema
+
+**Pitfalls**:
+- RETRIEVE_PAGE returns only metadata/properties, NOT body content; use FETCH_BLOCK_CONTENTS for page body
+- ARCHIVE_NOTION_PAGE is a soft-delete (sets archived=true), not permanent deletion
+- Broad searches can look incomplete unless has_more/next_cursor is fully paginated
+
+### 2. Query and Manage Databases
+
+**When to use**: User wants to query database rows, insert entries, or update records
+
+**Tool sequence**:
+1. `NOTION_SEARCH_NOTION_PAGE` - Find the database by name [Prerequisite]
+2. `NOTION_FETCH_DATABASE` - Inspect schema and properties [Prerequisite]
+3. `NOTION_QUERY_DATABASE` / `NOTION_QUERY_DATABASE_WITH_FILTER` - Query rows [Required]
+4. `NOTION_INSERT_ROW_DATABASE` - Add new entries [Optional]
+5. `NOTION_UPDATE_ROW_DATABASE` - Update existing entries [Optional]
+
+**Key parameters**:
+- `database_id`: Database ID (from search or URL)
+- `filter`: Filter object matching Notion filter syntax
+- `sorts`: Array of sort objects
+- `start_cursor`: Pagination cursor from previous response
+- `properties`: Property values matching database schema for inserts/updates
+
+**Pitfalls**:
+- 404 object_not_found usually means wrong database_id or the database is not shared with the integration
+- Results are paginated; ignoring has_more/next_cursor silently truncates reads
+- Schema mismatches or missing required properties cause 400 validation_error
+- Formula and read-only fields cannot be set via INSERT_ROW_DATABASE
+- Property names in filters must match schema exactly (case-sensitive)
+
+### 3. Manage Blocks and Page Content
+
+**When to use**: User wants to read, append, or modify content blocks in a page
+
+**Tool sequence**:
+1. `NOTION_FETCH_BLOCK_CONTENTS` - Read child blocks of a page [Required]
+2. `NOTION_ADD_MULTIPLE_PAGE_CONTENT` - Append blocks to a page [Optional]
+3. `NOTION_APPEND_TEXT_BLOCKS` - Append text-only blocks [Optional]
+4. `NOTION_REPLACE_PAGE_CONTENT` - Replace all page content [Optional]
+5. `NOTION_DELETE_BLOCK` - Remove a specific block [Optional]
+
+**Key parameters**:
+- `block_id` / `page_id`: Target page or block ID
+- `content_blocks`: Array of block objects (NOT child_blocks)
+- `text`: Plain text content for APPEND_TEXT_BLOCKS
+
+**Pitfalls**:
+- Use `content_blocks` parameter, NOT `child_blocks` -- the latter fails validation
+- ADD_MULTIPLE_PAGE_CONTENT fails on archived pages; unarchive via UPDATE_PAGE first
+- Created blocks are in response.data.results; persist block IDs for later edits
+- DELETE_BLOCK is archival (archived=true), not permanent deletion
+
+### 4. Manage Database Schema
+
+**When to use**: User wants to create databases or modify their structure
+
+**Tool sequence**:
+1. `NOTION_FETCH_DATABASE` - Inspect current schema [Prerequisite]
+2. `NOTION_CREATE_DATABASE` - Create a new database [Optional]
+3. `NOTION_UPDATE_SCHEMA_DATABASE` - Modify database properties [Optional]
+
+**Key parameters**:
+- `parent_id`: Parent page ID for new databases
+- `title`: Database title
+- `properties`: Property definitions with types and options
+- `database_id`: Database ID for schema updates
+
+**Pitfalls**:
+- Cannot change property types via UPDATE_SCHEMA; must create new property and migrate data
+- Formula, rollup, and relation properties have complex configuration requirements
+
+### 5. Manage Users and Comments
+
+**When to use**: User wants to list workspace users or manage comments on pages
+
+**Tool sequence**:
+1. `NOTION_LIST_USERS` - List all workspace users [Optional]
+2. `NOTION_GET_ABOUT_ME` - Get current authenticated user [Optional]
+3. `NOTION_CREATE_COMMENT` - Add a comment to a page [Optional]
+4. `NOTION_FETCH_COMMENTS` - List comments on a page [Optional]
+
+**Key parameters**:
+- `page_id`: Page ID for comments (also called `discussion_id`)
+- `rich_text`: Comment content as rich text array
+
+**Pitfalls**:
+- Comments are linked to pages, not individual blocks
+- User IDs from LIST_USERS are needed for people-type property filters
+
+## Common Patterns
+
+### ID Resolution
+
+**Page/Database name -> ID**:
+```
+1. Call NOTION_SEARCH_NOTION_PAGE with query=name
+2. Paginate with has_more/next_cursor until found
+3. Extract id from matching result
+```
+
+**Database schema inspection**:
+```
+1. Call NOTION_FETCH_DATABASE with database_id
+2. Extract properties object for field names and types
+3. Use exact property names in queries and inserts
+```
+
+### Pagination
+
+- Set `page_size` for results per page (max 100)
+- Check response for `has_more` boolean
+- Pass `start_cursor` or `next_cursor` in next request
+- Continue until `has_more` is false
+
+### Notion Filter Syntax
+
+**Single filter**:
+```json
+{"property": "Status", "select": {"equals": "Done"}}
+```
+
+**Compound filter**:
+```json
+{"and": [
+  {"property": "Status", "select": {"equals": "In Progress"}},
+  {"property": "Assignee", "people": {"contains": "user-id"}}
+]}
+```
+
+## Known Pitfalls
+
+**Integration Sharing**:
+- Pages and databases must be shared with the Notion integration to be accessible
+- Title queries can return 0 when the item is not shared with the integration
+
+**Property Types**:
+- Property names are case-sensitive and must match schema exactly
+- Formula, rollup, and created_time fields are read-only
+- Select/multi-select values must match existing options unless creating new ones
+
+**Response Parsing**:
+- Response data may be nested under `data_preview` or `data.results`
+- Parse defensively with fallbacks for different nesting levels
+
+## Quick Reference
+
+| Task | Tool Slug | Key Params |
+|------|-----------|------------|
+| Search pages/databases | NOTION_SEARCH_NOTION_PAGE | query |
+| Create page | NOTION_CREATE_NOTION_PAGE | parent_id, properties |
+| Get page metadata | NOTION_RETRIEVE_PAGE | page_id |
+| Update page | NOTION_UPDATE_PAGE | page_id, properties |
+| Archive page | NOTION_ARCHIVE_NOTION_PAGE | page_id |
+| Duplicate page | NOTION_DUPLICATE_PAGE | page_id |
+| Get page blocks | NOTION_FETCH_BLOCK_CONTENTS | block_id |
+| Append blocks | NOTION_ADD_MULTIPLE_PAGE_CONTENT | page_id, content_blocks |
+| Append text | NOTION_APPEND_TEXT_BLOCKS | page_id, text |
+| Replace content | NOTION_REPLACE_PAGE_CONTENT | page_id, content_blocks |
+| Delete block | NOTION_DELETE_BLOCK | block_id |
+| Query database | NOTION_QUERY_DATABASE | database_id, filter, sorts |
+| Query with filter | NOTION_QUERY_DATABASE_WITH_FILTER | database_id, filter |
+| Insert row | NOTION_INSERT_ROW_DATABASE | database_id, properties |
+| Update row | NOTION_UPDATE_ROW_DATABASE | page_id, properties |
+| Get database schema | NOTION_FETCH_DATABASE | database_id |
+| Create database | NOTION_CREATE_DATABASE | parent_id, title, properties |
+| Update schema | NOTION_UPDATE_SCHEMA_DATABASE | database_id, properties |
+| List users | NOTION_LIST_USERS | (none) |
+| Create comment | NOTION_CREATE_COMMENT | page_id, rich_text |
+| List comments | NOTION_FETCH_COMMENTS | page_id |