HolyJS 2026 Agent Runtime Workshop

This folder contains runnable code samples that map to the workshop plan. The practical flow is built around an email triage agent that classifies incoming emails into task, event, or no_action and routes them to downstream agents.

Step 01 uses a direct OpenAI-compatible HTTP call with no SDK, and falls back to naive keyword routing when OPENAI_API_KEY is not set. Step 02 uses the Google ADK (@google/adk) to define reusable agents. Later steps reuse those same agents for orchestration, n8n, and wrapper-based production concerns.

Quick start

cd 2026-02-28-holyjs-workshop
node -v
npm install
npm run start:01

Mapping to the agenda

1. Теория про агентов. Что такое вообще
   • See src/runtime/README.md for the runtime model and architecture notes.
2. Разработка агента "на коленке"
   • Run npm run start:01 and open src/01-standalone/run.js.
3. Теория про SDK для агентов
   • See src/02-sdk/README.md for the Google ADK overview.
4. Разработка агента на SDK
   • Run npm run start:02 and open src/02-sdk/run.js.
5. Теория про оркестрацию. Что делать, когда агентов несколько
   • Run npm run start:03 and open src/03-orchestrator/run.js.
6. Разработка ещё одного-двух агентов и встраивание его в n8n
   • Run npm run start:04 and open src/04-n8n/README.md.
7. Теория общих правил оркестрации. Безопасность, мониторинг и т.д.
   • Run npm run start:05 and open src/05-security-observability/run.js.
8. Практика
   • Use src/examples/ for exercises and tasks.
9. Конец

Structure

• src/01-standalone/ raw LLM email triage over HTTP without an SDK
• src/02-sdk/ Google ADK agents: classify email, simulate task creation, simulate agenda item creation
• src/03-orchestrator/ monolithic single-file orchestration over the step 02 agents
• src/04-n8n/ n8n integration nodes for visual orchestration over the step 02 agents
• src/05-security-observability/ wrappers for prompt-injection checks and success/error observability
• src/examples/ practice tasks and sample emails
• src/runtime/ runtime model notes (kept for theory section)
  • Also includes the small OpenAI-compatible helper reused by step 01

Notes

• Step 01 uses a real model when OPENAI_API_KEY is set, otherwise it falls back to naive keyword routing.
• Step 02 stays on the ADK code path even without credentials by using a local keyword-based BaseLlm fallback.
• Step 02 also supports Gemini, OpenAI, or a local OpenAI-compatible endpoint via SDK_PROVIDER; see src/02-sdk/README.md.
• Step 03 is intentionally monolithic and reuses the step 02 agents from one file.
• Step 04 reuses the same step 02 agents in n8n nodes, plus the step 03 router as a shortcut node.
• Step 05 wraps existing agents for prompt-injection blocking and success/error monitoring instead of re-implementing routing.