omegaprompt 2.2.0__tar.gz → 2.2.1__tar.gz

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/CHANGELOG.md

@@ -1,5 +1,24 @@
 # Changelog
 
+## [2.2.1] - 2026-10-28
+
+- Fix neutral reasoning and output-budget parameters to use indices into the
+  configured axis lists. Prefer STANDARD/MEDIUM when allowed, otherwise the first
+  allowed value; expose the same neutral through parameter specifications.
+- Correct the 2.2.0 default baseline from unintended DEEP to STANDARD. Valid
+  singleton subsets no longer fail bounds checks, and reordered lists preserve
+  the preferred semantic value. Historical artifacts are not rewritten; fresh
+  runs can have different baseline scores, requests and calibration outcomes.
+- Verify actual offline ProviderRequests and the runtime, CLI and MCP subset
+  calibration paths, retaining manual-review HOLD escalation.
+- Describe `PerformanceMeasurement.noise_floor` as absolute fitness standard
+  deviation, and describe the composite Action as checking declared artifact/gap/
+  hard-gate criteria without asserting generalization.
+
+The direct comparison baseline for this patch is public **2.2.0**, not 2.1.2.
+No gate threshold, artifact schema or existing reference fixture was relaxed.
+See [reproduction and migration](docs/release/2.2.1.md).
+
 ## [2.2.0] - 2026-10-28
 
 - Invalidate evaluation caches when public context changes, provide explicit fresh evaluation and cache invalidation, and keep boundary-clamp verdicts distinct.

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: omegaprompt
-Version: 2.2.0
+Version: 2.2.1
 Summary: Provider-neutral prompt calibration engine with sensitivity-ranked meta-axes, hard-gate x soft-score fitness, walk-forward ship gates, guarded/expedition profiles, and explicit capability/risk reporting across Anthropic, OpenAI, Gemini, and local OpenAI-compatible backends. Optional sub-tools (mini-omega-lock, mini-antemortem-cli) distribute separately and plug in via the omegaprompt.preflight interface.
 Project-URL: Homepage, https://github.com/hibou04-ops/omegaprompt
 Project-URL: Repository, https://github.com/hibou04-ops/omegaprompt
@@ -257,6 +257,10 @@ Description-Content-Type: text/markdown
 [![MCP](https://img.shields.io/badge/MCP-server-blueviolet.svg)](#103-mcp-server-claude-code-cursor)
 [![Parent framework](https://img.shields.io/badge/framework-omega--lock-blueviolet.svg)](https://github.com/hibou04-ops/omega-lock)
 
+> **2.2.1 (2026-10-28): neutral parameters respect configured subsets.** Select STANDARD/MEDIUM by their positions in the allowed lists, falling back to the first allowed value. Public 2.2.0 unintentionally used DEEP for the default baseline and rejected valid singleton lists. Fresh baseline scores, costs and selections can change; historical artifacts are preserved. [Reproduction and migration](docs/release/2.2.1.md).
+
+Raw optimizer integers are indices into the configured `MetaAxisSpace` lists, not the global `REASONING_ORDINALS`/`OUTPUT_BUDGET_ORDINALS` helper mappings.
+
 > **2.2.0 (2026-10-28): evaluation integrity and bounded work.** Context changes invalidate memoized observations. `PromptTarget.evaluate(..., use_cache=False)` performs fresh calls; cache replay adds no independent observation. Parallel evaluation retains at most `max_workers` submitted results and preserves input order. Failed target/judge attempts and returned usage remain in target accounting; no partial success is cached. Explicit disjoint mode rejects overlapping IDs. [Evidence and migration](docs/release/2.2.0.md).
 
 Docs: **[Easy start](https://github.com/hibou04-ops/omegaprompt/blob/main/EASY_README.md)** · [Full reference](https://github.com/hibou04-ops/omegaprompt/blob/main/README.md) · [한국어](https://github.com/hibou04-ops/omegaprompt/blob/main/README_KR.md) · [쉬운 한국어](https://github.com/hibou04-ops/omegaprompt/blob/main/EASY_README_KR.md) · [Examples gallery](examples/) · [Claim ledger (trust evidence)](docs/claims/README_CLAIMS.generated.md)
@@ -286,9 +290,9 @@ Calibrate prompt configurations through omega-lock and inspect CalibrationArtifa
 Requires Python 3.11+. Installation needs internet.
 
 ```bash
-python -m pip install omegaprompt==2.2.0
+python -m pip install omegaprompt==2.2.1
 omegaprompt --help
-python -c "from urllib.request import urlretrieve; urlretrieve('https://raw.githubusercontent.com/hibou04-ops/omegaprompt/v2.2.0/examples/reference/reference_artifact.json', 'artifact.json')"
+python -c "from urllib.request import urlretrieve; urlretrieve('https://raw.githubusercontent.com/hibou04-ops/omegaprompt/v2.2.1/examples/reference/reference_artifact.json', 'artifact.json')"
 omegaprompt check-artifact artifact.json
 omegaprompt report artifact.json --format json
 ```
@@ -406,7 +410,7 @@ A prompt change is a code change. Gate it like one. The dedicated **`omegaprompt
 
 ```yaml
 # .github/workflows/prompt-audit.yml
-- uses: hibou04-ops/omegaprompt@v2.2.0
+- uses: hibou04-ops/omegaprompt@v2.2.1
   with:
     artifact: artifact.json          # a CalibrationArtifact you produced in a prior step
     format: json                     # machine-readable gate summary

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/README.md

@@ -11,6 +11,10 @@
 [![MCP](https://img.shields.io/badge/MCP-server-blueviolet.svg)](#103-mcp-server-claude-code-cursor)
 [![Parent framework](https://img.shields.io/badge/framework-omega--lock-blueviolet.svg)](https://github.com/hibou04-ops/omega-lock)
 
+> **2.2.1 (2026-10-28): neutral parameters respect configured subsets.** Select STANDARD/MEDIUM by their positions in the allowed lists, falling back to the first allowed value. Public 2.2.0 unintentionally used DEEP for the default baseline and rejected valid singleton lists. Fresh baseline scores, costs and selections can change; historical artifacts are preserved. [Reproduction and migration](docs/release/2.2.1.md).
+
+Raw optimizer integers are indices into the configured `MetaAxisSpace` lists, not the global `REASONING_ORDINALS`/`OUTPUT_BUDGET_ORDINALS` helper mappings.
+
 > **2.2.0 (2026-10-28): evaluation integrity and bounded work.** Context changes invalidate memoized observations. `PromptTarget.evaluate(..., use_cache=False)` performs fresh calls; cache replay adds no independent observation. Parallel evaluation retains at most `max_workers` submitted results and preserves input order. Failed target/judge attempts and returned usage remain in target accounting; no partial success is cached. Explicit disjoint mode rejects overlapping IDs. [Evidence and migration](docs/release/2.2.0.md).
 
 Docs: **[Easy start](https://github.com/hibou04-ops/omegaprompt/blob/main/EASY_README.md)** · [Full reference](https://github.com/hibou04-ops/omegaprompt/blob/main/README.md) · [한국어](https://github.com/hibou04-ops/omegaprompt/blob/main/README_KR.md) · [쉬운 한국어](https://github.com/hibou04-ops/omegaprompt/blob/main/EASY_README_KR.md) · [Examples gallery](examples/) · [Claim ledger (trust evidence)](docs/claims/README_CLAIMS.generated.md)
@@ -40,9 +44,9 @@ Calibrate prompt configurations through omega-lock and inspect CalibrationArtifa
 Requires Python 3.11+. Installation needs internet.
 
 ```bash
-python -m pip install omegaprompt==2.2.0
+python -m pip install omegaprompt==2.2.1
 omegaprompt --help
-python -c "from urllib.request import urlretrieve; urlretrieve('https://raw.githubusercontent.com/hibou04-ops/omegaprompt/v2.2.0/examples/reference/reference_artifact.json', 'artifact.json')"
+python -c "from urllib.request import urlretrieve; urlretrieve('https://raw.githubusercontent.com/hibou04-ops/omegaprompt/v2.2.1/examples/reference/reference_artifact.json', 'artifact.json')"
 omegaprompt check-artifact artifact.json
 omegaprompt report artifact.json --format json
 ```
@@ -160,7 +164,7 @@ A prompt change is a code change. Gate it like one. The dedicated **`omegaprompt
 
 ```yaml
 # .github/workflows/prompt-audit.yml
-- uses: hibou04-ops/omegaprompt@v2.2.0
+- uses: hibou04-ops/omegaprompt@v2.2.1
   with:
     artifact: artifact.json          # a CalibrationArtifact you produced in a prior step
     format: json                     # machine-readable gate summary

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/README_KR.md

@@ -11,6 +11,10 @@
 [![MCP](https://img.shields.io/badge/MCP-server-blueviolet.svg)](https://github.com/hibou04-ops/omegaprompt/blob/main/README.md#103-mcp-server-claude-code-cursor)
 [![Parent framework](https://img.shields.io/badge/framework-omega--lock-blueviolet.svg)](https://github.com/hibou04-ops/omega-lock)
 
+> **2.2.1 (2026-10-28): 부분 축의 중립값 교정.** reasoning/budget 중립값은 설정 목록에서 STANDARD/MEDIUM의 위치를 사용하며, 없으면 첫 허용값을 사용합니다. 공개 2.2.0에서 기본 STANDARD가 DEEP으로 해석되거나 유효한 단일값 목록이 실패하던 결함을 교정했습니다. 새 실행의 기준 점수·비용·선택 결과는 달라질 수 있으며 기존 artifact는 보존합니다. [재현과 호환성](docs/release/2.2.1.md).
+
+optimizer의 원시 정수는 설정한 `MetaAxisSpace` 목록의 인덱스입니다. 전역 `REASONING_ORDINALS`/`OUTPUT_BUDGET_ORDINALS` helper 값과 혼동하지 마세요.
+
 > **2.2.0 (2026-10-28): 평가 무결성과 대기 작업 상한.** 데이터·rubric·variants 등 평가 컨텍스트 변경 시 캐시를 무효화합니다. `PromptTarget.evaluate(..., use_cache=False)`는 실제 재측정이며 캐시 재생은 독립 관측으로 세지 않습니다. 병렬 대기 결과는 `max_workers`개 이하이고 입력 순서를 보존합니다. 실패한 target/judge 시도와 반환된 usage도 누적 계측하며 부분 성공은 캐시하지 않습니다. 명시적 disjoint 모드는 ID 중복을 입력 오류로 처리합니다. [검증 근거와 이행 안내](docs/release/2.2.0.md).
 
 문서: **[Easy start](https://github.com/hibou04-ops/omegaprompt/blob/main/EASY_README.md)** · [Full reference](https://github.com/hibou04-ops/omegaprompt/blob/main/README.md) · [한국어](https://github.com/hibou04-ops/omegaprompt/blob/main/README_KR.md) · [쉬운 한국어](https://github.com/hibou04-ops/omegaprompt/blob/main/EASY_README_KR.md) · [Examples gallery](examples/) · [Claim ledger (신뢰 근거)](docs/claims/README_CLAIMS.generated.md)
@@ -40,16 +44,16 @@ omega-lock 엔진으로 프롬프트 설정을 보정하고 CalibrationArtifact
 Python 3.11 이상에서 실행합니다. 설치에는 인터넷이 필요합니다.
 
 ```bash
-python -m pip install omegaprompt==2.2.0
+python -m pip install omegaprompt==2.2.1
 omegaprompt --help
-python -c "from urllib.request import urlretrieve; urlretrieve('https://raw.githubusercontent.com/hibou04-ops/omegaprompt/v2.2.0/examples/reference/reference_artifact.json', 'artifact.json')"
+python -c "from urllib.request import urlretrieve; urlretrieve('https://raw.githubusercontent.com/hibou04-ops/omegaprompt/v2.2.1/examples/reference/reference_artifact.json', 'artifact.json')"
 omegaprompt check-artifact artifact.json
 omegaprompt report artifact.json --format json
 ```
 
 artifact 다운로드는 인터넷이 필요하며 이후 검사는 오프라인입니다. 저장된 합성 artifact의 누락된 per-item 점수는 transfer 통과를 뜻하지 않습니다. 실제 보정은 provider 설정과 비용 확인이 필요합니다. omega-lock>=0.3.0,<0.4.0은 필수이며 두 mini 도구는 선택입니다.
 
-[실제 연결 방식과 입력·출력](https://github.com/hibou04-ops/omega-lock/blob/main/DOCKING.md) · [전체 안내](README.md). 두 mini 도구의 현재 검증 조합은 omegaprompt 2.2.0이며, 별도 배포 패키지입니다.
+[실제 연결 방식과 입력·출력](https://github.com/hibou04-ops/omega-lock/blob/main/DOCKING.md) · [전체 안내](README.md). 두 mini 도구의 직전 검증 조합은 omegaprompt 2.2.0이며, 별도 배포 패키지입니다.
 
 MCP는 해당 배포명의 `[mcp]` extras와 기존 서버 명령을 사용합니다. 기존 FastMCP 계약은 SDK `>=1.0.0,<2.0.0` 범위로 유지합니다.
 
@@ -158,7 +162,7 @@ prompt 변경은 코드 변경입니다. 코드처럼 gate하세요. 전용 **`o
 
 ```yaml
 # .github/workflows/prompt-audit.yml
-- uses: hibou04-ops/omegaprompt@v2.2.0
+- uses: hibou04-ops/omegaprompt@v2.2.1
   with:
     artifact: artifact.json          # a CalibrationArtifact you produced in a prior step
     format: json                     # machine-readable gate summary

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/examples/_demo_output.txt

@@ -14,7 +14,7 @@ status: OK
 ship_recommendation: ship
 release_approved: True
 strict_blocking_findings: 0
-normalized_hash: 1b16c2cf423915fda7ea14779034e0dcab873f751ef0db265d82d02491ce5563
+normalized_hash: 4aaceafed3b741f6d5151e963c9a7674cf1d046a67115ceffc9a4403bc834070
 
 ---- Deterministic metrics from the artifact ----
 neutral_fitness:    0.4250

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/examples/ci/ship-gate.yml

@@ -24,12 +24,12 @@ jobs:
       # the committed reference artifact path.
       - name: Ship gate
         id: ship_gate
-        uses: hibou04-ops/omegaprompt@v2.2.0
+        uses: hibou04-ops/omegaprompt@v2.2.1
         with:
           artifact: examples/reference/reference_artifact.json
           format: json
           require-generalization: "true"
-          version: "2.2.0"
+          version: "2.2.1"
 
       - name: Report verdict
         if: always()

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/examples/reference/golden_manifest.json

@@ -12,7 +12,7 @@
       "expected_ship_recommendation": "ship",
       "expected_validation_mode": "auto",
       "expected_integrity_classification": "release_approved",
-      "normalized_artifact_hash": "1b16c2cf423915fda7ea14779034e0dcab873f751ef0db265d82d02491ce5563",
+      "normalized_artifact_hash": "4aaceafed3b741f6d5151e963c9a7674cf1d046a67115ceffc9a4403bc834070",
       "exact_metrics_may_be_displayed": true,
       "diff_baseline_case_id": null,
       "expected_diff_regression": null
@@ -25,7 +25,7 @@
       "expected_ship_recommendation": "hold",
       "expected_validation_mode": "paired",
       "expected_integrity_classification": "valid_non_release",
-      "normalized_artifact_hash": "58c5bc817a31dae49a2b838092fcd6e5cb6ddbc54ba57971b46fd430c2cfae26",
+      "normalized_artifact_hash": "bfbfe826b087138480c27b79704388fad2ebd215d47b37312c95e5e88978b317",
       "exact_metrics_may_be_displayed": true,
       "diff_baseline_case_id": null,
       "expected_diff_regression": null
@@ -38,7 +38,7 @@
       "expected_ship_recommendation": "block",
       "expected_validation_mode": "auto",
       "expected_integrity_classification": "valid_non_release",
-      "normalized_artifact_hash": "d163da066f2d72d36a30f076cf49d348ad22aa0a78ad7b1ba299fea8b103bf2b",
+      "normalized_artifact_hash": "a9392e2cc1d6756b3bd42b5a59416d743a96bc4f02ef9ea4bf3f01d1262af905",
       "exact_metrics_may_be_displayed": true,
       "diff_baseline_case_id": null,
       "expected_diff_regression": null
@@ -51,7 +51,7 @@
       "expected_ship_recommendation": "experiment",
       "expected_validation_mode": "auto",
       "expected_integrity_classification": "valid_non_release",
-      "normalized_artifact_hash": "7ae47e07afa1f08f9adbeb707d8dcb56bb19d0f1575700ffd8fa551650ec0418",
+      "normalized_artifact_hash": "4004021f470476c24207d30788ba12f1916e07d483cc9b3f2795b9d5e507ae6b",
       "exact_metrics_may_be_displayed": true,
       "diff_baseline_case_id": null,
       "expected_diff_regression": null
@@ -64,7 +64,7 @@
       "expected_ship_recommendation": "ship",
       "expected_validation_mode": "auto",
       "expected_integrity_classification": "release_approved",
-      "normalized_artifact_hash": "2f5541032e24771efa10824081050ec7465e53ca13940f7eda49b6deea0ac7fd",
+      "normalized_artifact_hash": "f02722870563895304d76fd0668b2265d0a4022c7dbbc897f6d17ce10be3fa0a",
       "exact_metrics_may_be_displayed": true,
       "diff_baseline_case_id": "clean_ok_ship",
       "expected_diff_regression": true

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/examples/reference/reference_artifact.json

@@ -7,7 +7,7 @@
   "neutral_baseline_params": {
     "system_prompt_variant": 0,
     "few_shot_count": 0,
-    "reasoning_profile": 2,
+    "reasoning_profile": 1,
     "output_budget_bucket": 1,
     "response_schema_mode": 0,
     "tool_policy_variant": 0

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/examples/reference/reference_diff_regression.json

@@ -7,7 +7,7 @@
   "neutral_baseline_params": {
     "system_prompt_variant": 0,
     "few_shot_count": 0,
-    "reasoning_profile": 2,
+    "reasoning_profile": 1,
     "output_budget_bucket": 1,
     "response_schema_mode": 0,
     "tool_policy_variant": 0

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/examples/reference/reference_fail_hard_gates.json

@@ -7,7 +7,7 @@
   "neutral_baseline_params": {
     "system_prompt_variant": 0,
     "few_shot_count": 0,
-    "reasoning_profile": 2,
+    "reasoning_profile": 1,
     "output_budget_bucket": 1,
     "response_schema_mode": 0,
     "tool_policy_variant": 0

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/examples/reference/reference_fail_kc4_gate.json

@@ -7,7 +7,7 @@
   "neutral_baseline_params": {
     "system_prompt_variant": 0,
     "few_shot_count": 0,
-    "reasoning_profile": 2,
+    "reasoning_profile": 1,
     "output_budget_bucket": 1,
     "response_schema_mode": 0,
     "tool_policy_variant": 0

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/examples/reference/reference_provider_degradation.json

@@ -7,7 +7,7 @@
   "neutral_baseline_params": {
     "system_prompt_variant": 0,
     "few_shot_count": 0,
-    "reasoning_profile": 2,
+    "reasoning_profile": 1,
     "output_budget_bucket": 1,
     "response_schema_mode": 0,
     "tool_policy_variant": 0

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "omegaprompt"
-version = "2.2.0"
+version = "2.2.1"
 description = "Provider-neutral prompt calibration engine with sensitivity-ranked meta-axes, hard-gate x soft-score fitness, walk-forward ship gates, guarded/expedition profiles, and explicit capability/risk reporting across Anthropic, OpenAI, Gemini, and local OpenAI-compatible backends. Optional sub-tools (mini-omega-lock, mini-antemortem-cli) distribute separately and plug in via the omegaprompt.preflight interface."
 readme = "README.md"
 requires-python = ">=3.11"

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/src/omegaprompt/__init__.py

@@ -114,7 +114,7 @@ from omegaprompt.runtime import (
     report,
 )
 
-__version__ = "2.2.0"
+__version__ = "2.2.1"
 
 __all__ = [
     # domain

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/src/omegaprompt/preflight/contracts.py

@@ -117,7 +117,7 @@ class PerformanceMeasurement(BaseModel):
     noise_floor: float = Field(
         default=0.0,
         ge=0.0,
-        description="Fitness variance from repeating an evaluation at fixed params.",
+        description="Absolute fitness standard deviation from repeating an evaluation at fixed params.",
     )
 
 

{omegaprompt-2.2.0 → omegaprompt-2.2.1}/src/omegaprompt/targets/prompt_target.py

@@ -13,9 +13,7 @@ from typing import Any
 
 from omegaprompt.domain.dataset import Dataset
 from omegaprompt.domain.enums import (
-    OUTPUT_BUDGET_ORDINALS,
     OutputBudgetBucket,
-    REASONING_ORDINALS,
     ReasoningProfile,
 )
 from omegaprompt.domain.judge import JudgeRubric
@@ -131,13 +129,20 @@ class PromptTarget:
         return {
             "system_prompt_variant": 0,
             "few_shot_count": self.space.few_shot_min,
-            "reasoning_profile": REASONING_ORDINALS[ReasoningProfile.STANDARD],
-            "output_budget_bucket": OUTPUT_BUDGET_ORDINALS[OutputBudgetBucket.MEDIUM],
+            "reasoning_profile": (
+                self.space.reasoning_profiles.index(ReasoningProfile.STANDARD)
+                if ReasoningProfile.STANDARD in self.space.reasoning_profiles else 0
+            ),
+            "output_budget_bucket": (
+                self.space.output_budgets.index(OutputBudgetBucket.MEDIUM)
+                if OutputBudgetBucket.MEDIUM in self.space.output_budgets else 0
+            ),
             "response_schema_mode": 0,
             "tool_policy_variant": 0,
         }
 
     def _param_specs_mini(self) -> list[_ParamSpecMini]:
+        neutral = self.neutral_params()
         return [
             _ParamSpecMini(
                 name="system_prompt_variant",
@@ -158,14 +163,14 @@ class PromptTarget:
                 dtype="int",
                 low=0,
                 high=len(self.space.reasoning_profiles) - 1,
-                neutral=min(1, len(self.space.reasoning_profiles) - 1),
+                neutral=neutral["reasoning_profile"],
             ),
             _ParamSpecMini(
                 name="output_budget_bucket",
                 dtype="int",
                 low=0,
                 high=len(self.space.output_budgets) - 1,
-                neutral=min(1, len(self.space.output_budgets) - 1),
+                neutral=neutral["output_budget_bucket"],
             ),
             _ParamSpecMini(
                 name="response_schema_mode",