chore: release mlbundle 0.2.1

AGENTS.md

@@ -5,7 +5,7 @@ inferred schema and model structure in one compressed `.mlbundle` file.
 
 It owns export, metadata-only inspection and validated loading. It is the
 training-side companion to MLSuite, whose runtime accepts `.mlbundle` uploads.
-MLSuite does not train models. The package version is 0.2.0; format 1 is not frozen.
+MLSuite does not train models. The package version is 0.2.1; format 1 is not frozen.
 Implementation claims require code and test proof.
 
 ## What must remain true

README.md

@@ -15,7 +15,7 @@
   <a href="https://mlbundle.eshkra.dev"><img src="https://img.shields.io/badge/website-303039?style=flat&amp;logo=astro&amp;logoColor=ff6a1a" alt="Website" /></a>
   <a href="https://docs.mlbundle.eshkra.dev"><img src="https://img.shields.io/badge/docs-303039?style=flat&amp;logo=readthedocs&amp;logoColor=ff6a1a" alt="Documentation" /></a>
   <a href="https://playground.mlbundle.eshkra.dev"><img src="https://img.shields.io/badge/playground-303039?style=flat&amp;logo=stackblitz&amp;logoColor=ff6a1a" alt="Playground" /></a>
-  <a href="https://pypi.org/project/mlbundle/"><img src="https://img.shields.io/badge/version-0.2.0-b73d00?style=flat&amp;logo=pypi&amp;logoColor=white&amp;labelColor=303039" alt="Version 0.2.0" /></a>
+  <a href="https://pypi.org/project/mlbundle/"><img src="https://img.shields.io/badge/version-0.2.1-b73d00?style=flat&amp;logo=pypi&amp;logoColor=white&amp;labelColor=303039" alt="Version 0.2.1" /></a>
   <a href="https://github.com/UlloaSP/mlbundle"><img src="https://img.shields.io/badge/source-303039?style=flat&amp;logo=github&amp;logoColor=white" alt="Source on GitHub" /></a>
   <a href="https://github.com/UlloaSP/mlbundle/actions/workflows/ci.yml"><img src="https://img.shields.io/github/actions/workflow/status/UlloaSP/mlbundle/ci.yml?branch=main&amp;style=flat&amp;label=CI&amp;labelColor=303039" alt="CI status" /></a>
   <a href="https://github.com/UlloaSP/mlbundle/blob/main/LICENSE"><img src="https://img.shields.io/badge/license-MIT-b73d00?style=flat&amp;labelColor=303039" alt="MIT license" /></a>
@@ -43,11 +43,11 @@ preserve the supported model without conversion.
 
 ## Installation
 
-MLbundle **0.2.0** requires Python **3.14** for export with the schema extra.
+MLbundle **0.2.1** requires Python **3.14** for export with the schema extra.
 The installation command for this release is:
 
 ```bash
-python -m pip install "mlbundle[sklearn,schema]==0.2.0"
+python -m pip install "mlbundle[sklearn,schema]==0.2.1"
 ```
 
 To use a local checkout, run `python -m pip install -e ".[sklearn,schema]"`

SECURITY.md

@@ -1,6 +1,6 @@
 # Security policy
 
-MLbundle 0.2.0 has not had an independent security review;
+MLbundle 0.2.1 has not had an independent security review;
 do not accept bundles from untrusted parties. Format version 1 is not frozen.
 It implements strict manifest validation, secure container reading/writing, metadata-only
 `inspect`, export/load, sklearn, XGBoost, CatBoost and ONNX adapters, local schema inference,

docs/publishing.md

@@ -1,6 +1,6 @@
 # Publishing to PyPI
 
-Version `0.2.0` is prepared by this checkout. Merging a version change into
+Version `0.2.1` is prepared by this checkout. Merging a version change into
 `main` automatically starts the PyPI workflow; no one needs to start a release
 run manually. Format version 1 is not frozen, so incompatible manifest/schema
 shapes require re-export. Valid a2 bundles remain compatible, as documented in
@@ -19,7 +19,7 @@ independent review; do not accept bundles from untrusted parties.
    tests on Ubuntu and Windows, and all frontend checks, tests and builds.
 4. If every check passes, Actions uploads the exact CI-built wheel and source
    archive to [mlbundle on PyPI](https://pypi.org/project/mlbundle/).
-   Install with `python -m pip install "mlbundle[sklearn,schema]==0.2.0"` on
+   Install with `python -m pip install "mlbundle[sklearn,schema]==0.2.1"` on
    Python 3.14, replacing the example version as needed.
 
 The GitHub `pypi_env` environment contains the project-scoped

docs/src/content/docs/getting-started/installation.md

@@ -1,23 +1,23 @@
 ---
 title: Installation
-description: Install mlbundle 0.2.0 with the extras for your model library, from PyPI or a local checkout, and check the Python and dependency ranges.
+description: Install mlbundle 0.2.1 with the extras for your model library, from PyPI or a local checkout, and check the Python and dependency ranges.
 ---
 
 mlbundle requires **Python 3.14 or newer**. Its only core dependency is Pydantic v2; every model
 library is an optional extra that is imported lazily, only when a bundle needs it.
 
-These pages document mlbundle 0.2.0.
+These pages document mlbundle 0.2.1.
 
 ```bash
-python -m pip install "mlbundle[sklearn,schema]==0.2.0"
+python -m pip install "mlbundle[sklearn,schema]==0.2.1"
 ```
 
 Replace `sklearn` with the extra for your model library; the extras are listed
 [below](#choose-the-extras).
 
 ## From a local checkout
 
-To work from source, or before 0.2.0 is available from your package index, install a checkout of
+To work from source, or before 0.2.1 is available from your package index, install a checkout of
 the repository:
 
 ```bash
@@ -66,7 +66,7 @@ from importlib.metadata import version
 
 import mlbundle
 
-print(version("mlbundle"))  # 0.2.0
+print(version("mlbundle"))  # 0.2.1
 print(mlbundle.FORMAT_NAME, mlbundle.FORMAT_VERSION)  # mlbundle 1
 ```
 

docs/src/content/docs/guides/cli.md

@@ -22,7 +22,7 @@ Metadata and integrity validated; model claims NOT verified.
 Format: mlbundle 1
 Bundle id: aa6f6b39-574a-499b-a0ff-de4cdd334050
 Created: 2026-10-19T11:17:51.949658+00:00
-Writer: "0.2.0"
+Writer: "0.2.1"
 Model: sklearn "DecisionTreeClassifier" (classifier)
 Counts: features=2, outputs=1, classes=[2]
 Features (declared): ["temperature", "humidity"]
@@ -34,7 +34,7 @@ Members:
   schema.json: 361 bytes sha256=ea62a6a7502ae8ea30b45b61bb18d8e31bd2ec55f026d9fbf0c05ef42d01ff25
   structure.json: 526 bytes sha256=66204da26781a4e2a800a1f3ed246ab1a43e8c97ced73d06a2640f897eeda03c
 Training: {"python": "3.14.7", "implementation": "CPython", "os": "Windows", "os_version": "11", "architecture": "AMD64"}
-Packages: {"catboost": "1.2.10", "mlbundle": "0.2.0", "mlschema": "0.3.0", ...}
+Packages: {"catboost": "1.2.10", "mlbundle": "0.2.1", "mlschema": "0.3.0", ...}
 ONNX environment: null
 Present: schema=yes, structure=yes, samples=no
 User metadata: {}

docs/src/content/docs/guides/inspect-and-load.md

@@ -27,7 +27,7 @@ from mlbundle import inspect
 
 view = inspect("trained.mlbundle")
 manifest = view.manifest
-print(manifest.format_version, manifest.writer_version)  # 1 0.2.0
+print(manifest.format_version, manifest.writer_version)  # 1 0.2.1
 print(manifest.model.n_features, manifest.model.n_outputs)  # 2 1
 print([member.value for member in view.members])
 # ['manifest.json', 'model.skops', 'schema.json', 'structure.json']

landing/src/data/example.json

@@ -1,8 +1,8 @@
 {
-  "version": "0.2.0",
+  "version": "0.2.1",
   "code": "import pandas as pd\nfrom sklearn.tree import DecisionTreeClassifier\n\nfrom mlbundle import export, load\n\ntraining = pd.DataFrame(\n    [[0.0, 1.0, \"cold\"], [1.0, 0.0, \"warm\"]] * 2,\n    columns=[\"temperature\", \"humidity\", \"target\"],\n)\nfeatures = training.drop(columns=\"target\")\nmodel = DecisionTreeClassifier(random_state=7)\nmodel.fit(features, training[\"target\"])\n\nexport(\n    model,\n    \"trained.mlbundle\",\n    training_data=training,\n    target_names=[\"target\"],\n)\n\nbundle = load(\"trained.mlbundle\")\nrecord = {\"temperature\": 0.0, \"humidity\": 1.0}\nprediction = bundle.predict_record(record)[0]\nprint(prediction.predicted_label, prediction.probabilities)\n",
   "printed": "cold (1.0, 0.0)",
-  "bundleBytes": 9227,
+  "bundleBytes": 9233,
   "members": [
     {
       "name": "manifest.json",
@@ -12,7 +12,7 @@
     {
       "name": "model.skops",
       "size": 15011,
-      "sha256": "13257b547ae5a12c2bc5cea20afaf68d1a36d6073c8f77f1ae676979c40375d6"
+      "sha256": "b332c1e060939f16a2f300797c6ee1625e99d684f215e45ba8772fd7dfee5da2"
     },
     {
       "name": "schema.json",
@@ -28,9 +28,9 @@
   "manifest": {
     "format_name": "mlbundle",
     "format_version": "1",
-    "bundle_id": "a9e717a9-b071-4d23-bbd1-6deedc47e6b5",
-    "created_at": "2026-10-19T11:17:33.114088Z",
-    "writer_version": "0.2.0",
+    "bundle_id": "869eb397-2dce-44b4-b0d0-b6a85d757b0b",
+    "created_at": "2026-10-19T13:27:32.304933Z",
+    "writer_version": "0.2.1",
     "model": {
       "estimator_class": "DecisionTreeClassifier",
       "kind": "classifier",
@@ -90,7 +90,7 @@
       "architecture": "AMD64",
       "packages": {
         "catboost": "1.2.10",
-        "mlbundle": "0.2.0",
+        "mlbundle": "0.2.1",
         "mlschema": "0.3.0",
         "numpy": "2.5.4",
         "onnx": "1.23.0",
@@ -105,7 +105,7 @@
     },
     "integrity": {
       "model.skops": {
-        "sha256": "13257b547ae5a12c2bc5cea20afaf68d1a36d6073c8f77f1ae676979c40375d6",
+        "sha256": "b332c1e060939f16a2f300797c6ee1625e99d684f215e45ba8772fd7dfee5da2",
         "size": 15011
       },
       "schema.json": {
@@ -159,7 +159,7 @@
   },
   "inspect": {
     "exit": 0,
-    "stdout": "Metadata and integrity validated; model claims NOT verified.\nFormat: mlbundle 1\nBundle id: a9e717a9-b071-4d23-bbd1-6deedc47e6b5\nCreated: 2026-10-19T11:17:33.114088+00:00\nWriter: \"0.2.0\"\nModel: sklearn \"DecisionTreeClassifier\" (classifier)\nCounts: features=2, outputs=1, classes=[2]\nFeatures (model): [\"temperature\", \"humidity\"]\nClass labels (output-major): [[\"cold\", \"warm\"]]\nTarget names (output-major): [\"target\"]\nMembers:\n  manifest.json: validated (no self-integrity record)\n  model.skops: 15011 bytes sha256=13257b547ae5a12c2bc5cea20afaf68d1a36d6073c8f77f1ae676979c40375d6\n  schema.json: 388 bytes sha256=e770ad53af881faf368c686de86b2b00f0a64a1444e241985643349bf7b0bba4\n  structure.json: 511 bytes sha256=8cf658be94b826ba4f81f1a78ed2c8e5107fce656d212dc888bd8621a9068baa\nTraining: {\"python\": \"3.14.7\", \"implementation\": \"CPython\", \"os\": \"Windows\", \"os_version\": \"11\", \"architecture\": \"AMD64\"}\nPackages: {\"catboost\": \"1.2.10\", \"mlbundle\": \"0.2.0\", \"mlschema\": \"0.3.0\", \"numpy\": \"2.5.4\", \"onnx\": \"1.23.0\", \"onnxruntime\": \"1.30.0\", \"pandas\": \"3.0.6\", \"scikit-learn\": \"1.9.1\", \"scipy\": \"1.18.1\", \"skops\": \"0.16.0\", \"xgboost\": \"3.4.1\"}\nONNX environment: null\nPresent: schema=yes, structure=yes, samples=no\nUser metadata: {}",
+    "stdout": "Metadata and integrity validated; model claims NOT verified.\nFormat: mlbundle 1\nBundle id: 869eb397-2dce-44b4-b0d0-b6a85d757b0b\nCreated: 2026-10-19T13:27:32.304933+00:00\nWriter: \"0.2.1\"\nModel: sklearn \"DecisionTreeClassifier\" (classifier)\nCounts: features=2, outputs=1, classes=[2]\nFeatures (model): [\"temperature\", \"humidity\"]\nClass labels (output-major): [[\"cold\", \"warm\"]]\nTarget names (output-major): [\"target\"]\nMembers:\n  manifest.json: validated (no self-integrity record)\n  model.skops: 15011 bytes sha256=b332c1e060939f16a2f300797c6ee1625e99d684f215e45ba8772fd7dfee5da2\n  schema.json: 388 bytes sha256=e770ad53af881faf368c686de86b2b00f0a64a1444e241985643349bf7b0bba4\n  structure.json: 511 bytes sha256=8cf658be94b826ba4f81f1a78ed2c8e5107fce656d212dc888bd8621a9068baa\nTraining: {\"python\": \"3.14.7\", \"implementation\": \"CPython\", \"os\": \"Windows\", \"os_version\": \"11\", \"architecture\": \"AMD64\"}\nPackages: {\"catboost\": \"1.2.10\", \"mlbundle\": \"0.2.1\", \"mlschema\": \"0.3.0\", \"numpy\": \"2.5.4\", \"onnx\": \"1.23.0\", \"onnxruntime\": \"1.30.0\", \"pandas\": \"3.0.6\", \"scikit-learn\": \"1.9.1\", \"scipy\": \"1.18.1\", \"skops\": \"0.16.0\", \"xgboost\": \"3.4.1\"}\nONNX environment: null\nPresent: schema=yes, structure=yes, samples=no\nUser metadata: {}",
     "stderr": ""
   },
   "verify": {
@@ -169,7 +169,7 @@
   },
   "editedInspect": {
     "exit": 0,
-    "stdout": "Metadata and integrity validated; model claims NOT verified.\nFormat: mlbundle 1\nBundle id: a9e717a9-b071-4d23-bbd1-6deedc47e6b5\nCreated: 2026-10-19T11:17:33.114088+00:00\nWriter: \"0.2.0\"\nModel: sklearn \"DecisionTreeClassifier\" (classifier)\nCounts: features=2, outputs=1, classes=[2]\nFeatures (model): [\"temperature\", \"humidity\"]\nClass labels (output-major): [[\"cold\", \"warm\"]]\nTarget names (output-major): [\"target\"]\nMembers:\n  manifest.json: validated (no self-integrity record)\n  model.skops: 15011 bytes sha256=13257b547ae5a12c2bc5cea20afaf68d1a36d6073c8f77f1ae676979c40375d6\n  schema.json: 388 bytes sha256=e770ad53af881faf368c686de86b2b00f0a64a1444e241985643349bf7b0bba4\n  structure.json: 511 bytes sha256=8cf658be94b826ba4f81f1a78ed2c8e5107fce656d212dc888bd8621a9068baa\nTraining: {\"python\": \"3.14.7\", \"implementation\": \"CPython\", \"os\": \"Windows\", \"os_version\": \"11\", \"architecture\": \"AMD64\"}\nPackages: {\"catboost\": \"1.2.10\", \"mlbundle\": \"0.2.0\", \"mlschema\": \"0.3.0\", \"numpy\": \"2.5.4\", \"onnx\": \"1.23.0\", \"onnxruntime\": \"1.30.0\", \"pandas\": \"3.0.6\", \"scikit-learn\": \"1.9.1\", \"scipy\": \"1.18.1\", \"skops\": \"0.16.0\", \"xgboost\": \"3.4.1\"}\nONNX environment: null\nPresent: schema=yes, structure=yes, samples=no\nUser metadata: {}",
+    "stdout": "Metadata and integrity validated; model claims NOT verified.\nFormat: mlbundle 1\nBundle id: 869eb397-2dce-44b4-b0d0-b6a85d757b0b\nCreated: 2026-10-19T13:27:32.304933+00:00\nWriter: \"0.2.1\"\nModel: sklearn \"DecisionTreeClassifier\" (classifier)\nCounts: features=2, outputs=1, classes=[2]\nFeatures (model): [\"temperature\", \"humidity\"]\nClass labels (output-major): [[\"cold\", \"warm\"]]\nTarget names (output-major): [\"target\"]\nMembers:\n  manifest.json: validated (no self-integrity record)\n  model.skops: 15011 bytes sha256=b332c1e060939f16a2f300797c6ee1625e99d684f215e45ba8772fd7dfee5da2\n  schema.json: 388 bytes sha256=e770ad53af881faf368c686de86b2b00f0a64a1444e241985643349bf7b0bba4\n  structure.json: 511 bytes sha256=8cf658be94b826ba4f81f1a78ed2c8e5107fce656d212dc888bd8621a9068baa\nTraining: {\"python\": \"3.14.7\", \"implementation\": \"CPython\", \"os\": \"Windows\", \"os_version\": \"11\", \"architecture\": \"AMD64\"}\nPackages: {\"catboost\": \"1.2.10\", \"mlbundle\": \"0.2.1\", \"mlschema\": \"0.3.0\", \"numpy\": \"2.5.4\", \"onnx\": \"1.23.0\", \"onnxruntime\": \"1.30.0\", \"pandas\": \"3.0.6\", \"scikit-learn\": \"1.9.1\", \"scipy\": \"1.18.1\", \"skops\": \"0.16.0\", \"xgboost\": \"3.4.1\"}\nONNX environment: null\nPresent: schema=yes, structure=yes, samples=no\nUser metadata: {}",
     "stderr": ""
   },
   "editedVerify": {

playground/README.md

@@ -8,7 +8,7 @@ Intended host: <https://playground.mlbundle.eshkra.dev/>. Documentation: <https:
 
 Python runs in your browser on [Pyodide](https://pyodide.org/), inside a dedicated Web Worker. There is no server that executes code, and neither your code nor your files are uploaded anywhere.
 
-- The build bundles this repository's `src/mlbundle` (every `.py` file plus the `state_fields.json` package data) into the worker, with distribution metadata whose version is read from `pyproject.toml`. The page always runs the mlbundle source it was built from, at version 0.2.0 today.
+- The build bundles this repository's `src/mlbundle` (every `.py` file plus the `state_fields.json` package data) into the worker, with distribution metadata whose version is read from `pyproject.toml`. The page always runs the mlbundle source it was built from, at version 0.2.1 today.
 - Each run executes your code as written, in a fresh namespace and an empty working directory. Every `*.mlbundle` the code writes into that directory (at most 8 per run) is passed to `mlbundle.inspect`, then to `mlbundle.load`, which deserializes it through the closed sklearn adapter and recomputes model claims and structure. Only then is it marked **Verified**. Loading is done automatically only for bundles your own code just wrote.
 - **Download .mlbundle** saves the exact bytes mlbundle wrote.
 - **Inspect a .mlbundle file** (the upload icon) reads a local file in the browser and runs `mlbundle.inspect` on it. It never calls `load` and never deserializes the model, so uploaded files show **Inspected only** and their structure is withheld: structure is only trustworthy once `load` has recomputed it. Files that mlbundle rejects show **Rejected** with mlbundle's error.

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "mlbundle"
-version = "0.2.0"
+version = "0.2.1"
 description = "Pickle-free, verifiable model bundles with a manifest, schema and structure."
 readme = "README.md"
 license = "MIT"

scripts/package_smoke.py

@@ -52,7 +52,7 @@ def _base(path: Path) -> None:
 
     import mlbundle
 
-    assert mlbundle.__version__ == "0.2.0"
+    assert mlbundle.__version__ == "0.2.1"
     view = mlbundle.inspect(path)
     assert view.manifest.model.library == "sklearn"
     assert "cannot attest" in _console("--help")