v1.6.1: fix live design crash (llm_call telemetry to stdlib logger)

CHANGELOG.md

@@ -5,6 +5,14 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [1.6.1] - 2026-07-08
+
+Hotfix for a crash in the flagship command, found by the July 2026 product audit.
+
+### Fixed
+
+- **`cloudwright design` no longer crashes on a live API key.** The per-call telemetry line passed structlog-style keyword arguments (`model=`, `tokens=`, ...) to a stdlib logger, so the CLI (which enables root INFO logging) raised `TypeError: Logger._log() got an unexpected keyword argument 'model'` right after the model responded and tokens were billed, producing no spec. The line now uses `%`-style stdlib formatting in both the Anthropic and OpenAI providers. Latent since v1.1.0: the test suite leaves the root logger at WARNING, so the line silently never emitted and never crashed under pytest. Regression test `test_llm_telemetry_logging.py` runs both providers under `configure_logging()`. As a side effect, per-call LLM telemetry (model, latency, tokens, cache hits) now actually emits.
+
 ## [1.6.0] - 2026-06-16
 
 This release closes the defensibility + relevance gaps from the June 2026 product audit:

packages/cli/cloudwright_cli/__init__.py

@@ -1 +1 @@
-__version__ = "1.6.0"
+__version__ = "1.6.1"

packages/core/cloudwright/__init__.py

@@ -17,7 +17,7 @@
     ValidationResult,
 )
 
-__version__ = "1.6.0"
+__version__ = "1.6.1"
 
 __all__ = [
     "Alternative",

packages/core/cloudwright/llm/anthropic.py

@@ -190,13 +190,14 @@ def _call(
                     "cached_tokens": cache_read,
                     "cache_creation_tokens": cache_write,
                 }
+                # stdlib logger: kwargs-style fields crash Logger._log, use %-style
                 log.info(
-                    "llm_call",
-                    model=model,
-                    duration_ms=round((time.perf_counter() - start) * 1000),
-                    tokens=usage["input_tokens"] + usage["output_tokens"],
-                    cache_read=cache_read,
-                    cache_write=cache_write,
+                    "llm_call model=%s duration_ms=%s tokens=%s cache_read=%s cache_write=%s",
+                    model,
+                    round((time.perf_counter() - start) * 1000),
+                    usage["input_tokens"] + usage["output_tokens"],
+                    cache_read,
+                    cache_write,
                 )
                 return response.content[0].text, usage
             except _RETRYABLE:

packages/core/cloudwright/llm/openai.py

@@ -217,12 +217,13 @@ def _call(
                     "output_tokens": response.usage.completion_tokens,
                     "cached_tokens": cached,
                 }
+                # stdlib logger: kwargs-style fields crash Logger._log, use %-style
                 log.info(
-                    "llm_call",
-                    model=model,
-                    duration_ms=round((time.perf_counter() - start) * 1000),
-                    tokens=usage["input_tokens"] + usage["output_tokens"],
-                    cached=cached,
+                    "llm_call model=%s duration_ms=%s tokens=%s cached=%s",
+                    model,
+                    round((time.perf_counter() - start) * 1000),
+                    usage["input_tokens"] + usage["output_tokens"],
+                    cached,
                 )
                 return content, usage
             except _RETRYABLE:

packages/core/pyproject.toml

@@ -29,10 +29,10 @@ dependencies = [
 ]
 
 [project.optional-dependencies]
-cli = ["cloudwright-ai-cli==1.6.0"]
-web = ["cloudwright-ai-cli==1.6.0", "cloudwright-ai-web==1.6.0"]
-mcp = ["cloudwright-ai-mcp==1.6.0"]
-all = ["cloudwright-ai-cli==1.6.0", "cloudwright-ai-web==1.6.0", "cloudwright-ai-mcp==1.6.0", "databricks-sdk>=0.38.0"]
+cli = ["cloudwright-ai-cli==1.6.1"]
+web = ["cloudwright-ai-cli==1.6.1", "cloudwright-ai-web==1.6.1"]
+mcp = ["cloudwright-ai-mcp==1.6.1"]
+all = ["cloudwright-ai-cli==1.6.1", "cloudwright-ai-web==1.6.1", "cloudwright-ai-mcp==1.6.1", "databricks-sdk>=0.38.0"]
 pdf = ["weasyprint", "markdown2"]
 databricks = ["databricks-sdk>=0.38.0"]
 live-import = [

packages/core/tests/test_llm_telemetry_logging.py

@@ -0,0 +1,74 @@
+"""Regression: the llm_call telemetry line must survive configure_logging().
+
+The CLI enables root INFO via configure_logging() on every invocation. Passing
+structlog-style kwargs to the stdlib logger returned by get_logger() raised
+TypeError: Logger._log() got an unexpected keyword argument 'model'
+AFTER the SDK call returned, so every live design/modify billed tokens and then
+crashed. pytest masked it because root stays at WARNING and info() short-circuits.
+These tests pin the fix by running the providers with root at INFO.
+"""
+
+from __future__ import annotations
+
+import logging as stdlib_logging
+from unittest.mock import MagicMock
+
+import cloudwright.logging as cw_logging
+import pytest
+from cloudwright.llm.anthropic import AnthropicLLM
+from cloudwright.llm.openai import OpenAILLM
+
+
+@pytest.fixture()
+def info_level_logging():
+    root = stdlib_logging.getLogger()
+    saved_handlers = root.handlers[:]
+    saved_level = root.level
+    saved_flag = cw_logging._configured
+    cw_logging._configured = False
+    cw_logging.configure_logging()
+    yield
+    root.handlers[:] = saved_handlers
+    root.setLevel(saved_level)
+    cw_logging._configured = saved_flag
+
+
+def _anthropic_response(text="ok"):
+    response = MagicMock()
+    response.content = [MagicMock(text=text)]
+    response.usage.input_tokens = 10
+    response.usage.output_tokens = 5
+    response.usage.cache_read_input_tokens = 0
+    response.usage.cache_creation_input_tokens = 0
+    return response
+
+
+def _openai_response(text="ok"):
+    response = MagicMock()
+    response.choices = [MagicMock(message=MagicMock(content=text))]
+    response.usage.prompt_tokens = 10
+    response.usage.completion_tokens = 5
+    response.usage.prompt_tokens_details = MagicMock(cached_tokens=0)
+    return response
+
+
+def test_anthropic_call_survives_configure_logging(info_level_logging):
+    llm = AnthropicLLM(api_key="test")
+    llm.client = MagicMock()
+    llm.client.messages.create.return_value = _anthropic_response("result")
+
+    text, usage = llm.generate([{"role": "user", "content": "hi"}], "system")
+
+    assert text == "result"
+    assert usage["input_tokens"] == 10
+
+
+def test_openai_call_survives_configure_logging(info_level_logging):
+    llm = OpenAILLM(api_key="test")
+    llm.client = MagicMock()
+    llm.client.chat.completions.create.return_value = _openai_response("result")
+
+    text, usage = llm.generate([{"role": "user", "content": "hi"}], "system")
+
+    assert text == "result"
+    assert usage["input_tokens"] == 10

packages/mcp/cloudwright_mcp/__init__.py

@@ -1 +1 @@
-__version__ = "1.6.0"
+__version__ = "1.6.1"

packages/web/cloudwright_web/__init__.py

@@ -1,6 +1,6 @@
 """Cloudwright Web — FastAPI backend for architecture intelligence."""
 
-__version__ = "1.6.0"
+__version__ = "1.6.1"
 
 
 def __getattr__(name: str):

server.json

@@ -2,7 +2,7 @@
   "$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json",
   "name": "io.github.xmpuspus/cloudwright",
   "description": "Natural-language cloud architecture: deployable Terraform/Pulumi, cost, compliance control mapping.",
-  "version": "1.6.0",
+  "version": "1.6.1",
   "repository": {
     "url": "https://github.com/xmpuspus/cloudwright",
     "source": "github"
@@ -12,7 +12,7 @@
       "registryType": "pypi",
       "registryBaseUrl": "https://pypi.org",
       "identifier": "cloudwright-ai-mcp",
-      "version": "1.6.0",
+      "version": "1.6.1",
       "transport": {
         "type": "stdio"
       }