feat(trigger-dev): GitHub AI agent demo (Hookdeck + Trigger.dev)

[leggetter]

Summary

Polishes the trigger-dev/github-ai-agent demo: interactive setup, Hookdeck + GitHub webhook fixes, Trigger.dev Production deploy + env sync, transformation + task hardening, docs.

Notes

• Draft PRs still emit pull_request opened / synchronize events — the AI review task runs the same unless we filter pull_request.draft in code.

Test plan

• npm run setup / npm run deploy from demo folder
• Issue labeler / PR review / push → Slack as applicable

Made with Cursor

[leggetter]

AI Review Summary

Summary

This PR adds a comprehensive GitHub AI agent demo showcasing Hookdeck + Trigger.dev integration. The demo includes interactive setup scripts, webhook handling, AI-powered automation tasks (PR reviews, issue labeling, deployment summaries), and documentation covering two different architectural patterns for webhook routing.

Key Observations

Strengths:

• Comprehensive automation: The setup scripts handle end-to-end configuration (Trigger.dev deployment, Hookdeck resources, GitHub webhooks) with proper ordering and idempotency
• Real-world use cases: AI-powered PR reviews, issue labeling, and deployment summaries are genuinely useful automation examples
• Multiple patterns: Shows both fan-out (single entry point) and direct routing approaches with clear trade-offs
• Security-focused: Three-layer verification chain (GitHub HMAC → Hookdeck verification → task-level validation)
• Production-ready: Proper error handling, environment variable management, and deployment strategies

Potential Concerns:

• File size: The .plan/ file is 932 lines and appears to be build documentation rather than runtime code - consider if this belongs in the repo
• Complexity: While comprehensive, the setup requires 6+ external services and API keys, which may be a barrier for quick experimentation
• Production-only: Hardcoded to production environment limits development workflow flexibility
• Draft PR behavior: The note about draft PRs still triggering reviews should be addressed in code rather than left as documentation

Suggestions for Improvement

1. Reduce complexity for first-time users: Consider a simplified "quick start" path with fewer dependencies (maybe just one task type) alongside the full demo
2. Environment flexibility: Allow development environment usage with appropriate configuration switches
3. Filter draft PRs: Add the draft PR check mentioned in the description: if (payload.pull_request?.draft) return;
4. Optimize file structure: Move the large .plan/ file to documentation or remove it if it's just build notes
5. Add validation: Include basic validation in setup scripts to verify API keys work before proceeding with full setup

The demo effectively showcases both platforms' capabilities and provides a solid foundation for developers to build GitHub automation workflows.

---

Generated by Hookdeck + Trigger.dev

[leggetter]

AI Review Summary

Summary

This PR adds a comprehensive GitHub AI agent demo showcasing the integration between Hookdeck (webhook routing) and Trigger.dev (task execution). The demo includes three AI-powered GitHub automation tasks: PR code review summaries, issue labeling, and deployment notifications to Slack, with two different architectural patterns for connecting the services.

Key Observations

Strengths:

• Excellent documentation and developer experience with complete automation scripts and clear setup instructions
• Well-architected comparison of two integration patterns (fan-out vs. Hookdeck routing) with clear trade-offs
• Comprehensive security model with three-layer verification chain
• Production-ready configuration with proper environment variable management and idempotent setup scripts
• Real-world utility - the tasks solve actual developer workflow problems

Areas of Concern:

• File size and complexity: The build plan document is 932 lines, which may indicate over-engineering for a demo
• Production-only deployment: Hardcoded to Production environment limits development flexibility and could be risky for demo purposes
• Multiple external dependencies: Requires accounts/keys for 5+ services (Hookdeck, Trigger.dev, GitHub, Anthropic, Slack)
• Error handling gaps: Setup scripts use set -euo pipefail but may not gracefully handle partial failures during multi-step setup

Suggestions for Improvement

1. Consider dev environment support: Add a --dev flag or separate dev configuration to allow safer experimentation without affecting production resources

2. Add rollback capability: Include cleanup scripts (npm run cleanup) to remove created webhooks and Hookdeck resources for easier iteration

3. Simplify initial experience: Consider a "quick start" mode with fewer integrations (maybe just PR reviews) to lower the barrier to entry

4. Add validation steps: Include health checks in setup scripts to verify each service integration before proceeding to the next

5. Break down the build plan: The 900+ line planning document suggests this might benefit from being split into multiple smaller, focused demos

The technical implementation looks solid and the automation is impressive, but consider whether the scope might be too broad for a single demo.

---

Generated by Hookdeck + Trigger.dev

[leggetter]

AI Review Summary

Summary

This PR introduces a comprehensive GitHub AI agent demo integrating Hookdeck and Trigger.dev. It includes setup automation, two different webhook routing patterns, and three AI-powered GitHub automation tasks (PR review, issue labeling, and deployment summaries to Slack).

Key Observations

Strengths:

• Excellent documentation: The extensive plan document and README provide clear architectural guidance and explain two distinct integration patterns
• Production-ready automation: Setup scripts handle the complete workflow from deployment to webhook registration
• Real-world utility: The three demo tasks (AI PR reviews, issue labeling, Slack summaries) solve actual developer pain points
• Security-conscious: Implements proper verification chain (GitHub HMAC → Hookdeck → Trigger.dev Bearer token → task-level checks)
• Educational value: Comparing Pattern A (fan-out in code) vs Pattern B (fan-out in Hookdeck) teaches architectural trade-offs

Potential Concerns:

• Complexity: The extensive documentation and dual patterns might overwhelm newcomers - consider a "quick start" path
• Resource duplication: Setup creates both patterns simultaneously, which could cause confusion or duplicate processing
• Production-only deployment: No local dev workflow shown, which limits development iteration
• External dependencies: Requires multiple API keys (GitHub, Anthropic, Slack, Hookdeck, Trigger.dev) which increases setup friction

Suggestions for Improvement

1. Simplify initial experience: Consider defaulting to one pattern in the main README, with the other as an "advanced" section
2. Add local development guidance: Show how developers can test changes before deploying to production
3. Include troubleshooting section: Common issues with webhook delivery, API rate limits, or authentication failures
4. Consider a minimal example: A single-task version that newcomers can get running quickly before exploring the full demo
5. Add cleanup instructions: How to tear down resources when done experimenting

The demo effectively showcases both platforms' capabilities and provides genuine developer value. The thorough documentation is impressive, though the complexity might benefit from a more gradual onboarding experience.

---

Generated by Hookdeck + Trigger.dev