Bump spec pin to v0.17.1 for proposal 0019

Proposal 0019 reframes llm-provider §8 as a catalog of wire-format
mappings, with the existing OpenAI-compatible body nested under
§8.1. Purely textual on the spec side, no behavioral change, no
new fixtures.

Refresh code and doc references to §8.X to match the new structure:
§8.1 → §8.1.1, §8.2 → §8.1.2, §8.3 → §8.1.3, §8.5.1 → §8.1.5.1,
§8.1.1 → §8.1.1.1.

Author: chris-colinsky

CHANGELOG.md

@@ -4,6 +4,12 @@ All notable changes to `openarmature-python` are documented in this file.
 
 The format follows [Keep a Changelog](https://keepachangelog.com/en/1.1.0/). The package follows [Semantic Versioning](https://semver.org/); pre-1.0 minor bumps may carry behavioral changes per [spec governance](https://github.com/LunarCommand/openarmature-spec/blob/main/GOVERNANCE.md).
 
+## [Unreleased]
+
+### Notes
+
+- **Pinned spec version bumped to v0.17.1.** Proposal 0019 (multi-provider wire-format extension) reframes llm-provider §8 as a catalog of wire-format mappings, with the existing OpenAI-compatible body nested under §8.1. Purely textual on the spec side — no behavioral change, no fixture changes. Code and doc references to §8.X updated to match the new structure (§8.1 → §8.1.1, §8.2 → §8.1.2, §8.3 → §8.1.3, §8.5.1 → §8.1.5.1, §8.1.1 → §8.1.1.1). All existing conformance fixtures continue to pass.
+
 ## [0.8.0] — 2026-05-23
 
 LLM-provider span payload and GenAI semconv release. Pinned spec

openarmature-spec

@@ -48,7 +48,7 @@ Repository = "https://github.com/LunarCommand/openarmature-python"
 Specification = "https://github.com/LunarCommand/openarmature-spec"
 
 [tool.openarmature]
-spec_version = "0.17.0"
+spec_version = "0.17.1"
 
 [dependency-groups]
 dev = [

pyproject.toml

@@ -1,4 +1,4 @@
 """OpenArmature: workflow framework for LLM pipelines and tool-calling agents."""
 
 __version__ = "0.8.0"
-__spec_version__ = "0.17.0"
+__spec_version__ = "0.17.1"

src/openarmature/__init__.py

@@ -1,5 +1,5 @@
-# Spec: realizes llm-provider §8 (concrete OpenAI provider) including
-# the §8.3 wire-error mapping table.
+# Spec: realizes llm-provider §8.1 (OpenAI-compatible wire-format mapping) including
+# the §8.1.3 wire-error mapping table.
 
 """OpenAI-compatible HTTPX-based provider.
 
@@ -389,7 +389,7 @@ async def _do_complete(
         return self._parse_response(cast("dict[str, Any]", payload_raw), schema_dict, schema_class)
 
     # ------------------------------------------------------------------
-    # Request building (spec §8.1)
+    # Request building (spec §8.1.1)
     # ------------------------------------------------------------------
 
     def _build_request_body(
@@ -433,7 +433,7 @@ def _build_request_body(
                 },
             }
         elif not include_response_format:
-            # On the fallback path the §8.5.1 contract is "response_format
+            # On the fallback path the §8.1.5.1 contract is "response_format
             # MUST NOT be on the wire." RuntimeConfig is extra="allow" so
             # a caller could pass response_format through via the extras
             # loop above; strip it here so the fallback contract holds
@@ -442,7 +442,7 @@ def _build_request_body(
         return body
 
     # ------------------------------------------------------------------
-    # Response parsing (spec §8.2)
+    # Response parsing (spec §8.1.2)
     # ------------------------------------------------------------------
 
     def _parse_response(
@@ -460,7 +460,7 @@ def _parse_response(
             raise ProviderInvalidResponse(f"response missing required fields: {exc}") from exc
         finish_reason: str = finish_reason_raw if isinstance(finish_reason_raw, str) else "error"
 
-        # Per §8.2 (and conformance fixture 005's
+        # Per §8.1.2 (and conformance fixture 005's
         # `function_call_legacy_finish_reason_mapping` case): the
         # legacy `finish_reason: "function_call"` value MUST be
         # normalized to the spec's `"tool_calls"`. This is a
@@ -717,13 +717,13 @@ def _augment_messages_with_schema_directive(
 
 
 def _message_to_wire(msg: Message) -> dict[str, Any]:
-    """Spec §8.1 request mapping for one message."""
+    """Spec §8.1.1 request mapping for one message."""
     if isinstance(msg, SystemMessage):
         return {"role": "system", "content": msg.content}
     if isinstance(msg, UserMessage):
-        # Dual-shape user content (§8.1): string maps directly; a
+        # Dual-shape user content (§8.1.1): string maps directly; a
         # content-block sequence maps to OpenAI's content-array form
-        # per §8.1.1.
+        # per §8.1.1.1.
         if isinstance(msg.content, str):
             return {"role": "user", "content": msg.content}
         return {
@@ -760,7 +760,7 @@ def _message_to_wire(msg: Message) -> dict[str, Any]:
     }
 
 
-# Spec §8.1.1: content-block to OpenAI content-array entry mapping.
+# Spec §8.1.1.1: content-block to OpenAI content-array entry mapping.
 # Both URL-referenced and inline-base64 image blocks go through
 # OpenAI's `image_url` entry shape; the inline case is expressed as
 # an RFC 2397 data: URI carrying media_type + base64_data. The
@@ -872,7 +872,7 @@ def classify_http_error(resp: httpx.Response) -> LlmProviderError:
     if status in (401, 403):
         return ProviderAuthentication(message or f"HTTP {status}")
     if status == 400:
-        # Spec §8.3: HTTP 400 bodies that indicate the bound model
+        # Spec §8.1.3: HTTP 400 bodies that indicate the bound model
         # rejected a content block map to provider_unsupported_content_block
         # rather than the generic provider_invalid_request. The
         # detection rule is implementation-defined.

src/openarmature/llm/providers/openai.py

@@ -484,7 +484,7 @@ class MockResponse(_AllowExtras):
     Permissive shape because the body's content mirrors OpenAI's wire
     format which is wide and evolving; modelling every field would
     duplicate the OpenAI schema. The ``llm-provider`` capability's
-    spec.md §8 is the authoritative shape.
+    spec.md §8.1 is the authoritative shape.
     """
 
     status: int | None = None

tests/conformance/harness/directives.py

@@ -4,7 +4,7 @@
 behavior in terms of OpenAI Chat Completions wire-format mock
 responses + expected ``Provider.complete()`` / ``Provider.ready()``
 outcomes. The harness drives the real :class:`OpenAIProvider` via
-``httpx.MockTransport`` so the wire-mapping path (spec §8) is
+``httpx.MockTransport`` so the wire-mapping path (spec §8.1) is
 exercised end-to-end — fixture 005 explicitly tests that mapping, so
 mocking at the Provider boundary would skip what we want to verify.
 

tests/conformance/test_llm_provider.py

@@ -9,7 +9,7 @@
 
 def test_package_versions() -> None:
     assert openarmature.__version__ == "0.8.0"
-    assert openarmature.__spec_version__ == "0.17.0"
+    assert openarmature.__spec_version__ == "0.17.1"
 
 
 def test_spec_version_matches_pyproject() -> None:

tests/test_smoke.py

@@ -273,7 +273,7 @@ def test_transient_categories_excludes_terminal_categories() -> None:
 
 
 # ---------------------------------------------------------------------------
-# classify_http_error — wire-mapping table (spec §8.3)
+# classify_http_error — wire-mapping table (spec §8.1.3)
 # ---------------------------------------------------------------------------