[Bug] Streaming Chunk Memory Accumulation

[unn-Known1]

Summary

For very long streaming responses, collected_chunks list can grow unbounded, potentially causing memory issues.

Location

src/gateway/main.py - chat_completions() function, streaming path

Current Behavior

In the generate() function, chunks are accumulated without limit.

Expected Behavior

Implement a limit on stored chunks or use streaming token extraction without full storage.

Priority

Low

References

• CWE-400: Resource Exhaustion