Bump version to 0.1.3; enhance logging with correlation_id for better error tracing and add request lifecycle middleware

docs/PRD.md

@@ -60,6 +60,7 @@ The proxy defaults to Opus because this deployment is intended for users whose N
 - As a Claude Code user, I can use tool calls and tool results.
 - As a multimodal user, I can send images through Claude-compatible image content blocks.
 - As a developer debugging setup, I can enable verbose logs without exposing tokens.
+- As a developer debugging issues, I can trace any error in the log file using a correlation_id that appears in both the client response and the log entry.
 
 ## Acceptance Criteria
 
@@ -70,6 +71,7 @@ The proxy defaults to Opus because this deployment is intended for users whose N
 - `--dev` startup reads current-directory `nexus-claude-api.local.json` instead of user config.
 - Default logs are written to `~/.config/nexus-claude-api/logs/nexus-claude-api-YYYY-MM-DD.log`.
 - `--dev` logs are written to current-directory `logs/nexus-claude-api-YYYY-MM-DD.log`.
+- All API errors (4xx, 5xx, network failures, client disconnects) are logged with correlation_id and duration.
 - Missing Nexus credentials fail fast with a clear error.
 - `GET /health` returns healthy status.
 - `GET /v1/models` returns the supported Claude models.

docs/REQUIREMENTS_DESIGN.md

@@ -29,7 +29,9 @@ nexus-claude-api/
       __main__.py
       cli.py
       config.py
+      diagnostics.py
       errors.py
+      logging_config.py
       models.py
       nexus_client.py
       server.py
@@ -212,6 +214,28 @@ Status mapping:
 - Nexus throttling: `429`
 - Nexus network/timeout: `502` or `504`
 - unexpected server error: `500`
+- client disconnected: `499` (logged only, not sent)
+
+## Observability
+
+All errors are logged to the daily log file with a `correlation_id` for tracing.
+
+Request lifecycle middleware logs:
+
+- `request_complete` with method, path, status, and duration_ms for every `/v1/messages` request.
+- `client_disconnected` when the client drops connection mid-request or mid-stream.
+- `unhandled_exception` with traceback for unexpected errors (returns 500).
+
+Error logging at point of origin:
+
+- `nexus_client_error` (WARNING): Nexus API returned an error (4xx/5xx) with error code, message, HTTP status, and request ID.
+- `nexus_botocore_error` (WARNING): Network/transport failure (timeout, DNS, connection reset).
+- `stream_nexus_error` (WARNING): Error during streaming response iteration.
+- `stream_client_disconnected` (INFO): Client closed connection during streaming.
+- `validation_error` (WARNING): Malformed request that failed pydantic validation.
+- `nexus_error` (WARNING): NexusClaudeError in non-streaming response path.
+
+Diagnostics summarize each request (model, stream flag, message count, content block types, tool info) without logging secrets, prompt text, or base64 data.
 
 ## Testing
 
@@ -234,6 +258,11 @@ Route tests:
 - `POST /v1/messages` non-stream
 - `POST /v1/messages` stream
 - `POST /v1/messages/count_tokens`
+- Validation errors are logged with correlation_id.
+- Nexus errors (non-stream) are logged with status and type.
+- Nexus errors (stream) are logged as `stream_nexus_error`.
+- Request completion is logged with duration_ms.
+- Unexpected exceptions return 500 and are logged.
 
 CLI tests: