FDE AI Workbench Plugin

Complete PagerDuty Professional Services consulting toolkit for Claude Code with proposal generation, script creation, customer analysis, and automated workflows.

• 4 Slash Commands: Script generation, IW action creation, customer analysis, health check remediation
• 5 Specialized Agents: Python, JavaScript, Terraform, Cloud Architecture, Documentation experts
• 65 Composable Skills: Automatic workflows with intelligent orchestration
• Dual Schema Registries: Snowflake (29 tables) + PagerDuty API (50+ endpoints)
• Essential Templates: Proposals, Python scripts, Lambda functions, Terraform configs
• 4 MCP Integrations: PagerDuty API, Snowflake, Atlassian (Jira/Confluence), Chrome DevTools

# Install from GitHub
claude plugin install github:PagerDuty/fde-ai-workbench-plugin

# Install specific version
claude plugin install github:PagerDuty/fde-ai-workbench-plugin@v1.0.0

# Clone and install locally
git clone https://github.com/PagerDuty/fde-ai-workbench-plugin.git
cd fde-ai-workbench-plugin
claude plugin install .

# Navigate to plugin directory
cd ~/.claude/plugins/fde-ai-wb

# Copy environment template
cp .env.example .env

# Edit and add your PagerDuty API token
nano .env

Set in .env:

PAGERDUTY_API_TOKEN=your_api_token_here

For Jira and Confluence integration:

# Add Atlassian MCP with browser authentication
claude mcp add --transport sse atlassian -s user https://mcp.atlassian.com/v1/sse

# Follow browser prompts to authenticate with your Atlassian account

If using Snowflake data queries:

# Edit Snowflake config file
nano ~/.config/snowflake/config.ini

Add your Snowflake credentials:

[default]
account = your_account
user = your_username
password = your_password
warehouse = your_warehouse
database = your_database
schema = your_schema

# List installed plugins
claude plugin list

# Test MCP connections
claude mcp list
claude mcp test pagerduty-api
claude mcp test atlassian

# View available commands
claude
> /help

All commands are namespaced with /fde-ai-wb: prefix.

Generate Python scripts for PagerDuty automation with python-pro agent expertise.

Usage:

/fde-ai-wb:create-script Create a user export script for Tesla Inc, hosted on AWS Lambda

Features:

• Standard Python template compliance
• Error handling and retry logic
• Configuration management
• Multiple hosting options (script, Lambda, Azure, Docker)
• Automatic documentation generation

Generated Output:

• Main Python script
• Configuration template (config.ini.example)
• Dependencies file (requirements.txt)
• README.md with usage instructions
• Hosting-specific files (if applicable)

Generate PagerDuty Incident Workflow Actions with automated testing.

Usage:

/fde-ai-wb:create-iw-action Create Slack action to send incident summaries

Features:

• JavaScript best practices with javascript-pro agent
• Automated unit testing with Docker
• Intelligent auto-fix for test failures
• Integration generation (if applicable)
• Lint validation

Generated Output:

• Action JavaScript file
• Test suite
• Integration files (if applicable)
• Test results with auto-fix report

Analyze customer requests with multi-source research.

Usage:

/fde-ai-wb:analyze-customer Analyze Tesla request for ServiceNow integration

Features:

• PagerDuty API feasibility check
• Documentation search
• Scripts library search
• Solution Gallery lookup
• Health check recommendations
• Optional proposal generation

Research Sources:

• PagerDuty OpenAPI schema
• Performance guides
• Existing scripts
• Solution Gallery (SRAs)
• Health Check catalog

Health check audit and remediation workflow.

Usage:

/fde-ai-wb:remediate

Features:

• 3-step workflow (audit → plan → execute)
• Dry-run support
• CSV bulk replacement mode
• Automated validation
• Core PagerDuty library integration

Supported Health Checks:

• HC-002: Schedules With Inactive Users
• HC-003: Escalation Policies With Inactive Users

These workflows trigger automatically based on natural language:

Trigger Keywords:

• "create proposal for [customer]"
• "generate proposal for [customer]"
• "I need a proposal for [customer]"

Example:

> Create a proposal for Tesla Inc about ServiceNow Integration, 40 hours, gold subscription
✅ Proposal generated automatically
✅ Multi-level Mermaid diagrams included
✅ Template validation with auto-fixes applied
✅ Saved to: ./Outputs/proposals/[PROPOSAL] Tesla Inc...

Trigger Keywords:

• "retrieve/get/fetch [data] from Snowflake"
• "I need [data] from account ID [number]"

Example:

> Get users from Snowflake for account 12345
✅ Searches existing queries
✅ Presents options to user
✅ Executes selected query
✅ Exports to CSV: ./Outputs/reports/AccountID_12345/users_2025-01-14.csv

Trigger Keywords:

• "create/generate report for [data]"
• "export [data] to CSV"

Example:

> Export incidents to CSV for account 12345
✅ Validates API + Snowflake feasibility
✅ Presents both options with trade-offs
✅ User selects method
✅ Generates report
✅ Exports to: ./Outputs/reports/AccountID_12345/incidents_2025-01-14.csv

registry/snowflake_schema_metadata.json provides:

• 29 table schemas with 200+ fields
• 10 common JOIN patterns
• Field usage statistics
• Query templates
• Anti-patterns documentation

Used for:

• Query validation before execution
• Auto-applying security filters (account_id, region, deleted_at)
• Suggesting proven JOIN patterns
• Warning about rarely-used fields

registry/pagerduty_openapiv3.json provides:

• Complete OpenAPI v3 specification
• 50+ endpoints documented
• Request/response schemas
• Parameter validation
• Rate limit documentation

Used for:

• API feasibility validation
• Field availability checking
• Filter support verification
• Operation limit identification
• Report method selection (API vs Snowflake)

Purpose: Direct PagerDuty REST API integration

Setup: Set PAGERDUTY_API_TOKEN in .env

Capabilities:

• Real-time incident, service, user management
• Webhook and integration management
• Schedule and escalation policy operations

Purpose: Data analytics and historical reporting

Setup: Configure ~/.config/snowflake/config.ini

Capabilities:

• Historical data analysis
• Complex aggregations and analytics
• Multi-table joins
• No record limits

Purpose: Ticket tracking and documentation publishing

Setup: Browser authentication (one-time)

claude mcp add --transport sse atlassian -s user https://mcp.atlassian.com/v1/sse

Capabilities:

• Create and update Jira tickets
• Publish to Confluence spaces
• Search documentation
• Track customer engagements

Purpose: Browser automation and E2E testing

Setup: No configuration needed

Capabilities:

• Automated browser testing
• Screenshot capture
• Network monitoring
• Integration debugging

# Start Claude in any project directory
cd ~/projects/tesla-engagement
claude

# Generate proposal
> Create a proposal for Tesla Inc about ServiceNow Integration, 40 hours
✅ Proposal: ./Outputs/proposals/[PROPOSAL] Tesla Inc...

# Create automation script
> /fde-ai-wb:create-script User export for Tesla, Lambda hosting
✅ Script: ./Outputs/scripts/Tesla_Inc/user_export/

# Analyze additional request
> /fde-ai-wb:analyze-customer Tesla wants incident reporting dashboard
✅ Analysis: API feasibility + recommendations + complexity

# Export user data
> Get users from Snowflake for account 500655

# Plugin automatically:
# 1. Searches Queries_library/ for existing queries
# 2. Presents 3 matching queries with recommendations
# 3. Waits for user selection
# 4. Executes chosen query
# 5. Exports to CSV

✅ Output: ./Outputs/reports/AccountID_500655/users_2025-01-14.csv

> /fde-ai-wb:remediate

# Interactive workflow guides through:
# 1. Select health check (HC-002, HC-003)
# 2. Run audit
# 3. Review plan
# 4. Choose dry-run or live execution
# 5. Apply fixes

✅ Schedules fixed
✅ Escalation policies corrected
✅ Audit trail saved

# Verify installation
claude plugin list

# Reinstall if needed
claude plugin install github:PagerDuty/fde-ai-workbench-plugin

# Test individual MCPs
claude mcp test pagerduty-api
claude mcp test atlassian

# Re-authenticate Atlassian
claude mcp remove atlassian
claude mcp add --transport sse atlassian -s user https://mcp.atlassian.com/v1/sse

# Verify command name includes namespace
/fde-ai-wb:create-script  # ✅ Correct
/create-script            # ❌ Wrong (missing namespace)

# List available commands
claude
> /help

# Verify token is set
cat ~/.claude/plugins/fde-ai-wb/.env

# Test API access
curl -H "Authorization: Token token=$PAGERDUTY_API_TOKEN" \
     https://api.pagerduty.com/users?limit=1

# Update to latest version
claude plugin update fde-ai-wb

# Check current version
claude plugin list

• ✅ Works from any directory - No need to clone workbench repo
• ✅ Install once, use everywhere - Consistent across all customer projects
• ✅ Automatic updates - Simple claude plugin update command
• ✅ Clean project structure - No large .claude/ folders in projects
• ✅ Team consistency - Everyone uses same version with same capabilities
• ✅ Intelligent validation - Dual schema registries prevent errors
• ✅ 4 MCP integrations - PagerDuty, Snowflake, Atlassian, Chrome DevTools
• ✅ 65 composable skills - Automatic workflows and progressive enhancement

This plugin uses the Composable Skills System following Anthropic's best practices:

• Small, focused skills (50-150 lines each)
• Orchestrated workflows (3-12 skills per workflow)
• Parallel execution where possible (40-70% speed improvement)
• Intelligent caching (24% faster on repeated operations)
• Continuous learning (execution logs, pattern recognition)

• Issues: https://github.com/PagerDuty/fde-ai-workbench-plugin/issues
• Documentation: See individual command files in commands/ directory
• Updates: Check CHANGELOG.md for version history

MIT License - See LICENSE file for details

Current version: 1.0.0

For complete version history, see CHANGELOG.md