homepage

Homepage knowledge agent

A voice agent that answers questions about LiveKit products. Knowledge about the Agents SDKs stays in the system prompt for low-latency answers, while other product knowledge is loaded on demand through a generated lookup_product tool.

Architecture

• agent.py is the composition root. It contains the immutable AgentConfig, Assistant, session setup, and server entrypoint.
• knowledge_base/ discovers one Markdown file per product and derives both the tool schema and lookup results from those files.
• prompts/ contains all authored agent language as Markdown templates.
• behaviors/ contains session event behavior and frontend integration.
• filters/ contains streaming voice-pipeline transformations.
• tests/unit/ is deterministic; tests/evals/ runs live behavioral evaluations.

The voice pipeline uses LiveKit Inference with Gemma 4 31B, Deepgram Nova-3, Fish Audio S2.1 Pro in expressive mode, the LiveKit turn detector, and ai-coustics voice isolation.

Run locally

Install the workspace dependencies and provide LiveKit Cloud credentials in ../.env or the environment:

uv sync --all-extras --dev   # from the repository root
uv run agent.py console

Use uv run agent.py dev to connect the agent to LiveKit Cloud for a frontend or telephony session.

Tests and evals

Install the repository's development dependencies, then run the fast unit suite:

python -m pytest

Run the live LLM-backed eval suite with:

python -m pytest -m evals

The evals require LIVEKIT_API_KEY and LIVEKIT_API_SECRET and cover tool routing, grounded facts, anti-hallucination behavior, inline knowledge, and multi-turn grounding.