Scaffold locally · serving

Chat agent

Knowledge, conversation memory, and the LLM answer are tables and computed columns.

Scaffold locally

pxt schema update app.py agent && pxt service update app.py agent

Same starter-kit files Cloud uses. Then pxt schema update and pxt service update. A generic app is pxt init plus pxt service example --out app.py. Extra DAGs live in starter-kit examples/.

Secrets

Set these on the database before calling model-backed routes: ANTHROPIC_API_KEY

Cloud routes

  • insert/api/ask→ agent/agent
  • insert/api/knowledge→ agent/knowledge
  • query/api/knowledge/search