PersonalProfile / CamDigitalProfile

Monorepo for Cameron Hammond’s recruiter-facing portfolio site plus an agentic AI backend (LangChain.js + Google Gemini + tools).

What's in the repo

• CamDigitalProfile/ — Static site (index.html, assets) with an embedded chat UI that talks to the agent API.
• agent-backend/ — Node.js (ESM) Express service: ReAct-style LangChain agent with calculator, web_search (Tavily), and embedding-based knowledge_base RAG over Cameron’s docs.

Run the agent backend locally

From the repo root:

cd agent-backend
npm install
npm run dev

The server listens on port 8787 by default (PORT overrides). Ensure a PersonalProfile/.env file (gitignored) exists at the repo root with:

• GEMINI_API_KEY — required for chat and document embeddings.
• TAVILY_API_KEY — required for live web search.

Optional env vars: see CamDigitalProfile/AI_Projects/docs/AGENT-BACKEND-RUNBOOK.md (GEMINI_MODEL, GEMINI_EMBEDDING_MODEL, AGENT_MAX_ITERATIONS, MODEL_TIMEOUT_MS, MEMORY_TURNS, etc.).

Tests and smoke

From agent-backend/:

• npm test — unit/regression checks (memory, vector ranking, agent graph with a fake tool-calling model, source guards against heuristic-only routing / keyword-only RAG).
• npm run test:integration — optional live Gemini embedding sanity check (requires GEMINI_API_KEY; exits 0 with a skip message if missing).
• npm run smoke — HTTP health check (GET /api/agent/health). Start the server first (npm run dev).

Use the site chat

Open CamDigitalProfile/index.html in a browser (e.g. Live Server). The UI posts to http://localhost:8787/api/agent/chat (configurable via window.AGENT_CHAT_ENDPOINT in the HTML).

Documentation

• Runbook & env: CamDigitalProfile/AI_Projects/docs/AGENT-BACKEND-RUNBOOK.md
• Product requirements: CamDigitalProfile/AI_Projects/aiDocs/custom-agent-creation-prd.md
• Rubric: CamDigitalProfile/AI_Projects/docs/AGENT-PROJECT-RUBRIC.md