You've already forked adk-python
mirror of
https://github.com/encounter/adk-python.git
synced 2026-07-09 18:19:28 -07:00
Jack Sun
8b0581f5c5
Add a new RateLimitPlugin that enforces global rate limiting across all
LLM models using a sliding window algorithm. The plugin blocks (waits)
when the rate limit is exceeded, ensuring requests are processed within
the configured limit.
Key features:
- Global rate limiting (default 15 QPM) across all models
- Sliding window algorithm for accurate tracking
- Automatic blocking when limit exceeded (no errors thrown)
- Thread-safe with asyncio locks
- Automatic cleanup of expired timestamps
Example usage:
```python
from google.adk.plugins import RateLimitPlugin
runner = Runner(
agents=[agent],
plugins=[RateLimitPlugin(max_requests_per_minute=15)]
)
```