Skip to content

quickstart.md

Latest commit

 

History

History
68 lines (45 loc) · 2.33 KB

quickstart.md

File metadata and controls

68 lines (45 loc) · 2.33 KB
Raw

Quickstart

This guide gets the maintained TypeScript stack running locally and points you at the first useful evaluation flow.

1. Install prerequisites

Required:

  • Node.js 18.17+ (20 is recommended)
  • npm

Optional:

  • Python 3.10+ if you are working on the legacy CrewAI service
  • A local LLM runtime if your chosen agent path depends on one

2. Install dependencies

Run from the repository root:

npm install

The root postinstall builds the shared packages used by the backend, frontend, and agent-service-api workspaces.

3. Create local environment files

cp backend/.env.example backend/.env
cp frontend/.env.example frontend/.env.local
cp agent-service-api/.env.example agent-service-api/.env

The defaults connect the local services together:

  • Frontend: http://localhost:3000
  • Backend: http://localhost:5100
  • Agent Service API: http://localhost:5003

The backend writes SQLite data to backend/data/agent-testing.db by default.

4. Start the app

npm run dev

This starts the backend, agent-service-api, and frontend together. The Python agent-service is not part of the default development path.

Open http://localhost:3000.

5. Run your first conversation

  1. Open LLM configs and add or verify the model/API configuration your agent will use.
  2. Open Agents and create or select the agent version you want to evaluate.
  3. Open Conversations and create a conversation script with realistic messages.
  4. Open Quick execute, select the agent and conversation, then execute.
  5. Open Jobs to track queue progress.
  6. Open Sessions to inspect the transcript, timing, token usage, and scoring signals.

Troubleshooting

  • If the frontend cannot reach the API, confirm frontend/.env.local has NEXT_PUBLIC_API_URL=http://localhost:5100.
  • If you created local env files before the default backend port changed, update backend/.env, frontend/.env.local, and agent-service-api/.env together.
  • If jobs stay pending, confirm agent-service-api/.env points BACKEND_URL at the backend and that npm run dev is still running.
  • If shared package types look stale, run npm run build:packages or restart npm run dev.
  • If you are working on CrewAI, start agent-service separately and verify backend/.env has the correct AGENT_SERVICE_URL.