Chat

• --assistant (required): Assistant name
• --message (required): The question or message
• --stream (optional flag): Enable streaming for faster perceived response

Workflow

1. Parse arguments. If assistant missing, run uv run scripts/list.py --json and ask the user to select.
2. If message missing, prompt user for their question.
3. Execute:

   bashCopy

   uv run scripts/chat.py \
     --assistant "assistant-name" \
     --message "user's question"

4. Display:
   • Assistant’s response
   • Citations table: citation number, source file, page numbers, position
   • Token usage statistics

Note: File URLs in citations are temporary signed links (~1 hour). They are not displayed in output.

Troubleshooting

Assistant not found — run list command, check for typos. No response or timeout — verify assistant has files uploaded and status is “ready” (not “indexing”). Empty or poor responses — assistant may lack relevant documents; suggest upload. PINECONE_API_KEY not set — export the variable or add to a .env file, then restart your IDE/agent session.