Configurable Two-Layer Retry Strategy

Configuration

```yaml

~/.hermes/config.yaml

agent:
max_api_retries: 10 # Outer loop: full API call retries
max_stream_retries: 10 # Inner loop: stream/connection retries
```

Retry behavior

Outer loop (max_api_retries, default 3):

• Respects Retry-After header from API response (capped at 5 min)
• Rate limits: exponential 5s * 2^n with +/-20% jitter, cap 5 min
• Other errors: exponential 2^n, cap 60s

Inner loop (max_stream_retries, default 2):

• Retries streaming request with fresh connection on ReadTimeout/connection drops
• Rebuilds the primary OpenAI client to purge dead pool connections

Links

• PR: feat(retries): configurable max_api_retries + max_stream_retries with smarter backoff #5571