feat: add platform matrix and adapter snapshot tests

.gitignore

@@ -3,3 +3,4 @@ dist/
 *.zip
 __pycache__/
 tests/tmp/
+tests/tmp_snapshot/

Makefile

@@ -1,6 +1,6 @@
 PYTHON ?= python3
 
-.PHONY: eval eval-suite package-check package-failure-check validate lint test clean
+.PHONY: eval eval-suite package-check package-failure-check snapshot-check validate lint test clean
 
 eval:
 	$(PYTHON) scripts/trigger_eval.py --description-file evals/improved_description.txt --cases evals/trigger_cases.json --baseline-description-file evals/baseline_description.txt
@@ -14,13 +14,16 @@ package-check:
 package-failure-check:
 	$(PYTHON) tests/verify_packager_failures.py
 
+snapshot-check:
+	$(PYTHON) tests/verify_adapter_snapshots.py
+
 validate:
 	$(PYTHON) scripts/validate_skill.py .
 
 lint:
 	$(PYTHON) scripts/lint_skill.py .
 
-test: eval eval-suite package-check package-failure-check validate lint
+test: eval eval-suite package-check package-failure-check snapshot-check validate lint
 
 clean:
-	rm -rf dist tests/tmp
+	rm -rf dist tests/tmp tests/tmp_snapshot

README.md

@@ -44,6 +44,14 @@ Or run everything together:
 make test
 ```
 
+## Results
+
+- `make test` passes locally
+- trigger eval: `0` false positives, `0` false negatives on the current regression set
+- eval suite: train / dev / holdout all pass
+- packaging validation: `openai`, `claude`, and `generic` targets pass contract checks
+- packaging failure fixtures: invalid metadata, invalid YAML, and unsupported targets fail as expected
+
 ## What It Does
 
 This project helps you create, refactor, evaluate, and package skills as durable capability bundles rather than one-off prompts.
@@ -198,7 +206,9 @@ This project is best for:
 - Examples: [examples/README.md](examples/README.md)
 - Evals: [evals/README.md](evals/README.md)
 - Packaging contracts: [references/packaging-contracts.md](references/packaging-contracts.md)
+- Platform capability matrix: [references/platform-capability-matrix.md](references/platform-capability-matrix.md)
 - Failure fixtures: [tests/fixtures](tests/fixtures)
+- Adapter snapshots: [tests/snapshots](tests/snapshots)
 - Evolution example: [examples/evolution-frontend-review/README.md](examples/evolution-frontend-review/README.md)
 
 ## License

docs/README.fr-FR.md

@@ -15,6 +15,13 @@ Il transforme des workflows bruts, des transcripts, des prompts, des notes et de
 2. Utilisez `yao-meta-skill` pour générer ou améliorer le paquet en mode scaffold, production ou library.
 3. Exécutez `context_sizer.py`, `trigger_eval.py` et `cross_packager.py` selon le besoin pour valider et exporter le résultat.
 
+## Results
+
+- `make test` passe actuellement
+- sur le jeu de régression courant, trigger eval a `0` faux positifs et `0` faux négatifs
+- les suites train / dev / holdout passent toutes
+- les contrats de packaging `openai`, `claude` et `generic` sont validés
+
 ## Ce que fait le projet
 
 Ce projet permet de créer, refactoriser, évaluer et empaqueter des skills comme des briques de capacité durables plutôt que comme des prompts ponctuels.

docs/README.ja-JP.md

@@ -15,6 +15,13 @@
 2. `yao-meta-skill` を使って scaffold、production、library のいずれかのモードでパッケージを生成または改善します。
 3. 必要に応じて `context_sizer.py`、`trigger_eval.py`、`cross_packager.py` を実行し、検証と出力を行います。
 
+## Results
+
+- 現在 `make test` は通過
+- 現在の回帰セットでは trigger eval が `0` false positives / `0` false negatives
+- train / dev / holdout の 3 層評価が通過
+- `openai`、`claude`、`generic` の packaging contract が通過
+
 ## 何をするものか
 
 このプロジェクトは、skill を単発の prompt ではなく、作成・改善・評価・配布できる持続的な能力パッケージとして扱えるようにします。

docs/README.ru-RU.md

@@ -15,6 +15,13 @@
 2. Используйте `yao-meta-skill`, чтобы сгенерировать или улучшить пакет в режиме scaffold, production или library.
 3. При необходимости запустите `context_sizer.py`, `trigger_eval.py` и `cross_packager.py`, чтобы проверить и экспортировать результат.
 
+## Results
+
+- `make test` сейчас проходит
+- на текущем regression-наборе trigger eval дает `0` false positives и `0` false negatives
+- все три набора train / dev / holdout проходят
+- packaging contracts для `openai`, `claude` и `generic` проходят проверку
+
 ## Что делает проект
 
 Этот проект помогает создавать, перерабатывать, оценивать и упаковывать skills как долговечные capability-пакеты, а не как одноразовые prompts.

docs/README.zh-CN.md

@@ -15,6 +15,13 @@
 2. 使用 `yao-meta-skill` 以 scaffold、production 或 library 模式生成或改进 skill 包。
 3. 按需要运行 `context_sizer.py`、`trigger_eval.py` 和 `cross_packager.py` 来检查并导出结果。
 
+## Results
+
+- 当前 `make test` 可通过
+- 当前回归集下 trigger eval 为 `0` 误触发、`0` 漏触发
+- train / dev / holdout 三层评测均通过
+- `openai`、`claude`、`generic` 三个目标的 packaging contract 校验通过
+
 ## 它能做什么
 
 这个项目帮助你把 skill 从一次性 prompt，升级成可创建、可重构、可评估、可打包的长期能力包。

references/platform-capability-matrix.md

@@ -0,0 +1,31 @@
+# Platform Capability Matrix
+
+This matrix describes the current packaging targets and their support level.
+
+| Target | Metadata Adapter | Output Contract | Snapshot Test | Notes |
+| --- | --- | --- | --- | --- |
+| `openai` | Yes | Yes | Yes | Generates `targets/openai/agents/openai.yaml` |
+| `claude` | Yes | Yes | Yes | Generates `targets/claude/README.md` plus adapter metadata |
+| `generic` | Yes | Yes | Yes | Uses neutral adapter metadata only |
+
+## Current Support Model
+
+- `openai`: strongest metadata adapter support
+- `claude`: lightweight compatibility adapter, behavior still relies mainly on neutral source files
+- `generic`: lowest-friction export for neutral Agent Skills consumers
+
+## Explicit Non-Goals
+
+This project does not yet implement:
+
+- runtime-specific behavior transforms
+- client SDK integration
+- provider-specific execution logic
+
+## Degradation Rule
+
+If a target cannot support a source feature directly:
+
+1. preserve the neutral source package
+2. emit a minimal adapter
+3. document the fallback in the target output

tests/snapshots/claude_adapter.json

@@ -0,0 +1,5 @@
+{
+  "platform": "claude",
+  "canonical_metadata": "agents/interface.yaml",
+  "required_generated_file": "targets/claude/README.md"
+}

tests/snapshots/generic_adapter.json

@@ -0,0 +1,5 @@
+{
+  "platform": "generic",
+  "canonical_metadata": "agents/interface.yaml",
+  "required_generated_file": "targets/generic/adapter.json"
+}

tests/snapshots/openai_adapter.json

@@ -0,0 +1,5 @@
+{
+  "platform": "openai",
+  "canonical_metadata": "agents/interface.yaml",
+  "required_generated_file": "targets/openai/agents/openai.yaml"
+}

tests/verify_adapter_snapshots.py

@@ -0,0 +1,62 @@
+#!/usr/bin/env python3
+import json
+import shutil
+import subprocess
+import sys
+from pathlib import Path
+
+
+ROOT = Path(__file__).resolve().parent.parent
+SCRIPT = ROOT / "scripts" / "cross_packager.py"
+EXPECTATIONS = ROOT / "evals" / "packaging_expectations.json"
+SNAPSHOTS = ROOT / "tests" / "snapshots"
+TMP = ROOT / "tests" / "tmp_snapshot"
+
+
+def main() -> None:
+    if TMP.exists():
+        shutil.rmtree(TMP)
+    proc = subprocess.run(
+        [
+            sys.executable,
+            str(SCRIPT),
+            str(ROOT),
+            "--platform",
+            "openai",
+            "--platform",
+            "claude",
+            "--platform",
+            "generic",
+            "--expectations",
+            str(EXPECTATIONS),
+            "--output-dir",
+            str(TMP),
+        ],
+        cwd=ROOT,
+        capture_output=True,
+        text=True,
+    )
+    if proc.returncode != 0:
+        print(proc.stdout)
+        print(proc.stderr)
+        raise SystemExit(proc.returncode)
+
+    failures = []
+    for name in ("openai", "claude", "generic"):
+        snapshot = json.loads((SNAPSHOTS / f"{name}_adapter.json").read_text(encoding="utf-8"))
+        adapter = json.loads((TMP / "targets" / name / "adapter.json").read_text(encoding="utf-8"))
+        if adapter.get("platform") != snapshot["platform"]:
+            failures.append(f"{name}: platform mismatch")
+        if adapter.get("canonical_metadata") != snapshot["canonical_metadata"]:
+            failures.append(f"{name}: canonical metadata mismatch")
+        if not (TMP / snapshot["required_generated_file"]).exists():
+            failures.append(f"{name}: missing generated file {snapshot['required_generated_file']}")
+
+    report = {"ok": not failures, "failures": failures}
+    print(json.dumps(report, ensure_ascii=False, indent=2))
+    if failures:
+        raise SystemExit(2)
+
+
+if __name__ == "__main__":
+    main()