pipecat-ai-whisker

What it is and what it does

Whisker is a low-level debugger built specifically for the Pipecat framework, which orchestrates multi-agent voice and multimodal AI systems. It attaches observers to your Pipecat workers to capture and stream frame events, job flows, and bus messages to a graphical interface where you can browse the entire system topology, inspect individual pipelines, filter and trace frames through processors, and save sessions for later replay. The debugger runs as a WebSocket server (default localhost:9090) that the UI connects to—either a hosted version via ngrok tunneling or a local Node.js development server.

You integrate Whisker by adding a WhiskerServer instance and per-worker WhiskerObserver objects to your Pipecat runner, or by using environment-variable-driven setup files to avoid modifying application code. For CI and headless recording, you can use WhiskerFile instead to write events directly to disk without opening a port. The package is designed as a trace-logging tool with a UI layer, letting you see exactly what every part of a distributed Pipecat system is doing at any moment.

Use it for:

• Debug message flow and frame routing through complex multi-worker Pipecat pipelines during development
• Inspect job queues and worker state to diagnose deadlocks or unexpected behavior in live systems
• Record and replay Pipecat sessions for post-mortem analysis or sharing debugging context with teammates
• Monitor frame processing in real-time to identify bottlenecks or processors consuming unexpected input
• Capture headless Pipecat runs in CI/testing environments by writing events to disk without UI overhead

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Whisker is a real-time debugger for Pipecat voice and multimodal AI applications that lets you inspect workers, pipelines, jobs, and message flow across a multi-agent system through a graphical UI.

Yes, if you are actively developing or debugging Pipecat applications. The package is actively maintained, has no known vulnerabilities, installs with low friction, and provides a purpose-built debugging interface for a framework that otherwise lacks built-in introspection. The permissive BSD-2-Clause license poses no restrictions. Install it as a development dependency; the overhead of observers is negligible for debugging but can be disabled in production.

Install

pipecat-ai-whisker on PyPI

pip install pipecat-ai-whisker

uv add pipecat-ai-whisker

poetry add pipecat-ai-whisker

Installing pipecat-ai-whisker

Before you install

Low friction install with a pure-Python wheel. Requires Python 3.11+ and an active Pipecat installation; the package itself has only six runtime dependencies (aiofiles, idna, loguru, msgpack, pipecat-ai, websockets). Actively maintained with recent releases.

License in practice

BSD-2-Clause is permissive and imposes minimal restrictions on use or redistribution; you can incorporate this into commercial or proprietary projects without licensing concerns.

Quickstart

pip install pipecat-ai-whisker

from pipecat_ai_whisker import WhiskerServer
from pipecat import PipelineWorker, WorkerRunner

whisker = WhiskerServer()
worker.add_observer(whisker.create_observer(worker))
await runner.add_workers(whisker, worker)
await runner.run()

Requires Python 3.11+, an installed Pipecat framework, and Node.js 20+ if running the UI locally. The hosted UI also requires ngrok for local tunneling.

Verify before relying

• Whether the package works with Pipecat versions other than the latest, or version constraints on pipecat-ai dependency
• Performance overhead of Whisker observers on frame throughput in production-scale pipelines
• Whether saved .whisk session files are forward/backward compatible across Whisker versions

Package facts

Evidence: pipecat_ai_whisker-2.0.0-py3-none-any.whl

Keywords: pipecat, debug, ai

Tags

More Artificial Intelligence packages

Further reading