predikit 0.6.2__tar.gz → 0.6.3__tar.gz

{predikit-0.6.2 → predikit-0.6.3}/CHANGELOG.md

@@ -1,141 +1,149 @@
-# Changelog
-
-All notable changes to this project will be documented in this file.
-
-The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
-
-## [Unreleased]
-
-## [0.6.2] - 2026-08-22
-
-### Added
-- `py.typed` marker, so type checkers actually consume the annotations. Every module is annotated and mypy-clean, and the package declares the `Typing :: Typed` classifier, but without the PEP 561 marker downstream consumers got no type information at all.
-
-### Changed
-- `from_snowflake()` imports `snowflake.ml.registry` inside the function body instead of at module level, matching `from_mlflow()` and the project's own rule for optional dependencies. The module-level `Registry = None` sentinel is gone.
-- CI lints and format-checks `examples/` alongside `src/` and `tests/`.
-
-### Removed
-- `uv.lock`. Nothing consumed it — CI installs with `pip install -e ".[dev]"` — and a library should not pin its consumers' transitive dependencies.
-- `.claude/settings.local.json` from version control. It holds per-developer machine paths.
-
-## [0.6.1] - 2026-08-15
-
-### Fixed
-- MCP tools now advertise an `outputSchema` and return `structuredContent`. The generated callable carried no return annotation, so FastMCP emitted `outputSchema: null` and every result came back as unstructured text. Output fields are named from the tool's `output_name` / `output_description`, and a `collect` ensemble declares each member's output.
-- Low-confidence results keep their `_confidence` and `_low_confidence` keys when served over MCP. The output model allows extra keys; a strict one validates them away silently.
-- `predikit serve` no longer disguises unrelated failures as a bad registry target. Module import and attribute lookup are handled separately, and errors raised while building or running the server propagate with their own traceback instead of being reported as `Could not load registry target`.
-- `coerce_value()` converts `True` / `False` to `1` / `0` for `int` fields. `isinstance(True, int)` is true in Python, so bools previously passed through unconverted.
-
-### Added
-- `--host` and `--port` options on `predikit serve`, plus matching `host` / `port` arguments on `create_mcp_server()`. The `streamable-http` transport was previously stuck on the SDK default of `127.0.0.1:8000`.
-- `ToolRegistry.items()`, `.names()`, `__iter__`, and `__len__` as the public way to enumerate a registry. `create_mcp_server()` used the private `_tools` / `_ensembles` attributes; `to_openai()` and `to_langchain()` now share the same accessor.
-- Python 3.13 in the CI matrix and the package classifiers.
-
-### Changed
-- `mcp` is now a dev dependency, so the MCP integration is tested against the real SDK. The previous tests substituted a fake `FastMCP` and reported full coverage of `mcp.py` while never exercising the SDK — which is how the missing output schema shipped.
-- Test coverage for `predikit serve`, which previously had none.
-
-## [0.6.0] - 2026-08-03
-
-### Added
-- MCP server integration via `create_mcp_server()` and `predikit serve`.
-- Optional `mcp` dependency for exposing `ToolRegistry` tools to MCP-compatible clients.
-
-## [0.5.2] - 2026-08-01
-
-### Fixed
-- Repaired the development lockfile so `uv` can resolve the pinned Colorama artifact.
-- Added repository-wide LF line-ending normalization so Ruff formatting is stable across platforms.
-
-### Added
-- Regression coverage for invalid Snowflake output methods and duplicate registry names.
-- CI package-build verification to catch release artifacts that cannot be built.
-
-## [0.5.1] - 2026-07-21
-
-### Added
-- Test coverage for the LangChain exporter (`test_exporters_langchain.py`) — previously untested
-- Warning when `confidence_threshold` is set on a classifier whose model has no `predict_proba` method, mirroring the existing regressor warning
-- Test coverage for both `confidence_threshold` construction-time warnings (regressor, and classifier without `predict_proba`)
-
-### Fixed
-- `to_langchain()` now wires up `ainvoke()` as the LangChain tool's `coroutine`, so async LangChain agents actually get async execution instead of silently falling back to sync
-
-## [0.5.0] - 2026-07-01
-
-### Fixed
-- Improved `ModelTool` input coercion for optional scalar annotations like `float | None`, `bool | None`, `int | None`, and `str | None`
-- Added test coverage for optional scalar string inputs to keep LLM-friendly coercion consistent across modern Pydantic schemas
-
-### Changed
-- Removed generated MLflow run artifacts from source control and ignored future `mlruns/` outputs
-- Cleaned package metadata and CLI text to avoid Unicode dash rendering issues in terminals and package indexes
-
-## [0.4.5] - 2026-07-01
-
-### Fixed
-- Added validation for `confidence_threshold` so values outside `0.0` to `1.0` fail at construction time
-- Added `ModelEnsemble` validation for duplicate `collect` outputs, mismatched aggregate output names, negative weights, and zero-total weighted strategies
-- Added `ToolRegistry` duplicate-name validation across tools and ensembles
-- Updated `ToolRegistry.get()` to retrieve both tools and ensembles with clearer missing-name errors
-- Cleaned up example lint issues and corrected stale `modelbridge[xgboost]` install text
-
-### Changed
-- Marked MLflow loader tests as opt-in integration tests so the default pytest suite stays fast and deterministic
-
-## [0.4.4] - 2026-06-24
-
-### Changed
-- README overhauled: added "Why predikit?" comparison table, "Works with" ecosystem section, GitHub star/fork badges, and a shipped-vs-planned roadmap
-- Quick start section replaces "30-second example" heading; `---` dividers added between major sections for better scanability
-- PyPI package description updated to a more descriptive one-liner
-- Keywords expanded: added `scikit-learn`, `tool-use`, `openai`, `langchain`, `mlops`, `ai-agents`, `pydantic`, `model-serving`
-- Added `Documentation` URL to `[project.urls]` in `pyproject.toml`
-- Added `Topic :: Software Development :: Libraries :: Python Modules` and `Typing :: Typed` classifiers
-
-## [0.4.2] - 2026-06-13
-
-### Changed
-- Redesigned PyPI/README hero: logo, centered tagline, and badges in a unified `<p align="center">` block
-- Tagline moved from a `##` heading to a proper descriptive paragraph
-- Badges converted to centered HTML `<img>` links for consistent rendering on PyPI
-- Quick code teaser repositioned directly below badges (before Table of Contents)
-- "Field naming rule" added to Table of Contents
-- `ainvoke()` added to `ModelTool` Core API reference table
-- `ModelEnsemble` Core API subsection added with constructor signature and full strategy table
-- Project Traffic / download badge moved to bottom of README
-- Development Status classifier bumped from `3 - Alpha` to `4 - Beta` in `pyproject.toml`
-- Removed CI test status badge from README
-
-## [0.4.1] - 2026-06-02
-
-### Added
-- `ruff` (lint + format) and `mypy` (type checking) configured in `pyproject.toml`
-- `.pre-commit-config.yaml` — ruff and mypy hooks run automatically before every commit
-- Lint CI job in GitHub Actions — runs `ruff check`, `ruff format --check`, and `mypy` on every push and PR
-- `CONTRIBUTING.md` — development setup, code style, and PR guidelines
-- `CHANGELOG.md`
-- `CLAUDE.md` — project context for Claude Code
-
-### Changed
-- Bumped version to `0.4.1`
-- Added `ruff>=0.4.0`, `mypy>=1.10`, and `pre-commit>=3.0` to `[dev]` extras
-
-## [0.4.0] - 2026-05-01
-
-### Added
-- `ModelEnsemble` with `weighted_mean` and `weighted_vote` strategies
-- `ainvoke()` async wrapper on `ModelTool` (runs blocking predict in a thread pool)
-- Verbose logging via `verbose=True` on `ModelTool`
-- Snowflake Model Registry loader (`from_snowflake`)
-- MLflow Model Registry loader (`from_mlflow`)
-- `predikit inspect` CLI command
-
-### Changed
-- Moved to `src/` layout with hatchling build backend
-- Upgraded Pydantic dependency to v2
-
-## [0.3.x] and earlier
-
-See [GitHub Releases](https://github.com/Tejas-TA/predikit/releases) for earlier history.
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [Unreleased]
+
+## [0.6.3] - 2026-09-13
+
+### Fixed
+- Corrected the MCP quick-start command: `predikit serve` requires both the `mcp` and `cli` extras. The MCP library API itself remains available with `predikit[mcp]` alone.
+
+### Changed
+- Polished the README installation guidance and documented MCP HTTP host/port options.
+
+## [0.6.2] - 2026-08-22
+
+### Added
+- `py.typed` marker, so type checkers actually consume the annotations. Every module is annotated and mypy-clean, and the package declares the `Typing :: Typed` classifier, but without the PEP 561 marker downstream consumers got no type information at all.
+
+### Changed
+- `from_snowflake()` imports `snowflake.ml.registry` inside the function body instead of at module level, matching `from_mlflow()` and the project's own rule for optional dependencies. The module-level `Registry = None` sentinel is gone.
+- CI lints and format-checks `examples/` alongside `src/` and `tests/`.
+
+### Removed
+- `uv.lock`. Nothing consumed it — CI installs with `pip install -e ".[dev]"` — and a library should not pin its consumers' transitive dependencies.
+- `.claude/settings.local.json` from version control. It holds per-developer machine paths.
+
+## [0.6.1] - 2026-08-15
+
+### Fixed
+- MCP tools now advertise an `outputSchema` and return `structuredContent`. The generated callable carried no return annotation, so FastMCP emitted `outputSchema: null` and every result came back as unstructured text. Output fields are named from the tool's `output_name` / `output_description`, and a `collect` ensemble declares each member's output.
+- Low-confidence results keep their `_confidence` and `_low_confidence` keys when served over MCP. The output model allows extra keys; a strict one validates them away silently.
+- `predikit serve` no longer disguises unrelated failures as a bad registry target. Module import and attribute lookup are handled separately, and errors raised while building or running the server propagate with their own traceback instead of being reported as `Could not load registry target`.
+- `coerce_value()` converts `True` / `False` to `1` / `0` for `int` fields. `isinstance(True, int)` is true in Python, so bools previously passed through unconverted.
+
+### Added
+- `--host` and `--port` options on `predikit serve`, plus matching `host` / `port` arguments on `create_mcp_server()`. The `streamable-http` transport was previously stuck on the SDK default of `127.0.0.1:8000`.
+- `ToolRegistry.items()`, `.names()`, `__iter__`, and `__len__` as the public way to enumerate a registry. `create_mcp_server()` used the private `_tools` / `_ensembles` attributes; `to_openai()` and `to_langchain()` now share the same accessor.
+- Python 3.13 in the CI matrix and the package classifiers.
+
+### Changed
+- `mcp` is now a dev dependency, so the MCP integration is tested against the real SDK. The previous tests substituted a fake `FastMCP` and reported full coverage of `mcp.py` while never exercising the SDK — which is how the missing output schema shipped.
+- Test coverage for `predikit serve`, which previously had none.
+
+## [0.6.0] - 2026-08-03
+
+### Added
+- MCP server integration via `create_mcp_server()` and `predikit serve`.
+- Optional `mcp` dependency for exposing `ToolRegistry` tools to MCP-compatible clients.
+
+## [0.5.2] - 2026-08-01
+
+### Fixed
+- Repaired the development lockfile so `uv` can resolve the pinned Colorama artifact.
+- Added repository-wide LF line-ending normalization so Ruff formatting is stable across platforms.
+
+### Added
+- Regression coverage for invalid Snowflake output methods and duplicate registry names.
+- CI package-build verification to catch release artifacts that cannot be built.
+
+## [0.5.1] - 2026-07-21
+
+### Added
+- Test coverage for the LangChain exporter (`test_exporters_langchain.py`) — previously untested
+- Warning when `confidence_threshold` is set on a classifier whose model has no `predict_proba` method, mirroring the existing regressor warning
+- Test coverage for both `confidence_threshold` construction-time warnings (regressor, and classifier without `predict_proba`)
+
+### Fixed
+- `to_langchain()` now wires up `ainvoke()` as the LangChain tool's `coroutine`, so async LangChain agents actually get async execution instead of silently falling back to sync
+
+## [0.5.0] - 2026-07-01
+
+### Fixed
+- Improved `ModelTool` input coercion for optional scalar annotations like `float | None`, `bool | None`, `int | None`, and `str | None`
+- Added test coverage for optional scalar string inputs to keep LLM-friendly coercion consistent across modern Pydantic schemas
+
+### Changed
+- Removed generated MLflow run artifacts from source control and ignored future `mlruns/` outputs
+- Cleaned package metadata and CLI text to avoid Unicode dash rendering issues in terminals and package indexes
+
+## [0.4.5] - 2026-07-01
+
+### Fixed
+- Added validation for `confidence_threshold` so values outside `0.0` to `1.0` fail at construction time
+- Added `ModelEnsemble` validation for duplicate `collect` outputs, mismatched aggregate output names, negative weights, and zero-total weighted strategies
+- Added `ToolRegistry` duplicate-name validation across tools and ensembles
+- Updated `ToolRegistry.get()` to retrieve both tools and ensembles with clearer missing-name errors
+- Cleaned up example lint issues and corrected stale `modelbridge[xgboost]` install text
+
+### Changed
+- Marked MLflow loader tests as opt-in integration tests so the default pytest suite stays fast and deterministic
+
+## [0.4.4] - 2026-06-24
+
+### Changed
+- README overhauled: added "Why predikit?" comparison table, "Works with" ecosystem section, GitHub star/fork badges, and a shipped-vs-planned roadmap
+- Quick start section replaces "30-second example" heading; `---` dividers added between major sections for better scanability
+- PyPI package description updated to a more descriptive one-liner
+- Keywords expanded: added `scikit-learn`, `tool-use`, `openai`, `langchain`, `mlops`, `ai-agents`, `pydantic`, `model-serving`
+- Added `Documentation` URL to `[project.urls]` in `pyproject.toml`
+- Added `Topic :: Software Development :: Libraries :: Python Modules` and `Typing :: Typed` classifiers
+
+## [0.4.2] - 2026-06-13
+
+### Changed
+- Redesigned PyPI/README hero: logo, centered tagline, and badges in a unified `<p align="center">` block
+- Tagline moved from a `##` heading to a proper descriptive paragraph
+- Badges converted to centered HTML `<img>` links for consistent rendering on PyPI
+- Quick code teaser repositioned directly below badges (before Table of Contents)
+- "Field naming rule" added to Table of Contents
+- `ainvoke()` added to `ModelTool` Core API reference table
+- `ModelEnsemble` Core API subsection added with constructor signature and full strategy table
+- Project Traffic / download badge moved to bottom of README
+- Development Status classifier bumped from `3 - Alpha` to `4 - Beta` in `pyproject.toml`
+- Removed CI test status badge from README
+
+## [0.4.1] - 2026-06-02
+
+### Added
+- `ruff` (lint + format) and `mypy` (type checking) configured in `pyproject.toml`
+- `.pre-commit-config.yaml` — ruff and mypy hooks run automatically before every commit
+- Lint CI job in GitHub Actions — runs `ruff check`, `ruff format --check`, and `mypy` on every push and PR
+- `CONTRIBUTING.md` — development setup, code style, and PR guidelines
+- `CHANGELOG.md`
+- `CLAUDE.md` — project context for Claude Code
+
+### Changed
+- Bumped version to `0.4.1`
+- Added `ruff>=0.4.0`, `mypy>=1.10`, and `pre-commit>=3.0` to `[dev]` extras
+
+## [0.4.0] - 2026-05-01
+
+### Added
+- `ModelEnsemble` with `weighted_mean` and `weighted_vote` strategies
+- `ainvoke()` async wrapper on `ModelTool` (runs blocking predict in a thread pool)
+- Verbose logging via `verbose=True` on `ModelTool`
+- Snowflake Model Registry loader (`from_snowflake`)
+- MLflow Model Registry loader (`from_mlflow`)
+- `predikit inspect` CLI command
+
+### Changed
+- Moved to `src/` layout with hatchling build backend
+- Upgraded Pydantic dependency to v2
+
+## [0.3.x] and earlier
+
+See [GitHub Releases](https://github.com/Tejas-TA/predikit/releases) for earlier history.

{predikit-0.6.2 → predikit-0.6.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: predikit
-Version: 0.6.2
+Version: 0.6.3
 Summary: Wrap any trained sklearn/XGBoost model as an LLM-callable tool: auto-generated JSON schemas, Pydantic I/O, OpenAI & LangChain ready, zero boilerplate.
 Project-URL: Homepage, https://github.com/Tejas-TA/predikit
 Project-URL: Repository, https://github.com/Tejas-TA/predikit
@@ -99,6 +99,8 @@ tool.invoke({"sqft": 2200})   # → {"price_usd": 370730}
 - [Core API](#core-api)
 - [Field naming rule](#field-naming-rule)
 - [Cookbook](#cookbook)
+- [Command line](#command-line)
+- [MCP server](#mcp-server)
 - [Roadmap](#roadmap)
 - [Contributing](#contributing)
 - [License](#license)
@@ -149,6 +151,8 @@ pip install predikit[xgboost]    # XGBoost support
 pip install predikit[langchain]  # LangChain StructuredTool export
 pip install predikit[mlflow]     # MLflow Model Registry loader
 pip install predikit[snowflake]  # Snowflake Model Registry loader
+pip install predikit[cli]        # `predikit inspect` and `predikit serve`
+pip install predikit[mcp,cli]    # serve a registry over MCP
 ```
 
 ---
@@ -333,14 +337,16 @@ agent = initialize_agent(tools=registry.to_langchain(), ...)
 Expose the same registry to MCP-compatible clients with the optional MCP extra:
 
 ```bash
-pip install predikit[mcp]
+pip install predikit[mcp,cli]
 predikit serve my_tools:registry
 ```
 
 `my_tools:registry` must point to a `ToolRegistry`, or to a zero-argument function
 that returns one. The default transport is `stdio`; use `--transport streamable-http`
-when serving over HTTP. MCP tools use the same Pydantic input validation and model
-execution as direct `invoke()` calls.
+when serving over HTTP, optionally with `--host` and `--port`. MCP tools advertise
+their output schemas and use the same Pydantic input validation and model execution
+as direct `invoke()` calls. For an in-process server instead of the CLI, install
+`predikit[mcp]` and call `create_mcp_server(registry)`.
 
 ### Bool inputs from an LLM
 

{predikit-0.6.2 → predikit-0.6.3}/README.md

@@ -42,6 +42,8 @@ tool.invoke({"sqft": 2200})   # → {"price_usd": 370730}
 - [Core API](#core-api)
 - [Field naming rule](#field-naming-rule)
 - [Cookbook](#cookbook)
+- [Command line](#command-line)
+- [MCP server](#mcp-server)
 - [Roadmap](#roadmap)
 - [Contributing](#contributing)
 - [License](#license)
@@ -92,6 +94,8 @@ pip install predikit[xgboost]    # XGBoost support
 pip install predikit[langchain]  # LangChain StructuredTool export
 pip install predikit[mlflow]     # MLflow Model Registry loader
 pip install predikit[snowflake]  # Snowflake Model Registry loader
+pip install predikit[cli]        # `predikit inspect` and `predikit serve`
+pip install predikit[mcp,cli]    # serve a registry over MCP
 ```
 
 ---
@@ -276,14 +280,16 @@ agent = initialize_agent(tools=registry.to_langchain(), ...)
 Expose the same registry to MCP-compatible clients with the optional MCP extra:
 
 ```bash
-pip install predikit[mcp]
+pip install predikit[mcp,cli]
 predikit serve my_tools:registry
 ```
 
 `my_tools:registry` must point to a `ToolRegistry`, or to a zero-argument function
 that returns one. The default transport is `stdio`; use `--transport streamable-http`
-when serving over HTTP. MCP tools use the same Pydantic input validation and model
-execution as direct `invoke()` calls.
+when serving over HTTP, optionally with `--host` and `--port`. MCP tools advertise
+their output schemas and use the same Pydantic input validation and model execution
+as direct `invoke()` calls. For an in-process server instead of the CLI, install
+`predikit[mcp]` and call `create_mcp_server(registry)`.
 
 ### Bool inputs from an LLM
 

{predikit-0.6.2 → predikit-0.6.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "predikit"
-version = "0.6.2"
+version = "0.6.3"
 description = "Wrap any trained sklearn/XGBoost model as an LLM-callable tool: auto-generated JSON schemas, Pydantic I/O, OpenAI & LangChain ready, zero boilerplate."
 readme = "README.md"
 license = {text = "MIT"}

{predikit-0.6.2 → predikit-0.6.3}/src/predikit/__init__.py

@@ -1,16 +1,16 @@
-"""Wrap fitted scikit-learn / XGBoost models as LLM-callable tools."""
-
-from .ensemble import ModelEnsemble
-from .exceptions import LowConfidenceError
-from .mcp import create_mcp_server
-from .registry import ToolRegistry
-from .tool import ModelTool
-
-__all__ = [
-    "LowConfidenceError",
-    "ModelEnsemble",
-    "ModelTool",
-    "ToolRegistry",
-    "create_mcp_server",
-]
-__version__ = "0.6.2"
+"""Wrap fitted scikit-learn / XGBoost models as LLM-callable tools."""
+
+from .ensemble import ModelEnsemble
+from .exceptions import LowConfidenceError
+from .mcp import create_mcp_server
+from .registry import ToolRegistry
+from .tool import ModelTool
+
+__all__ = [
+    "LowConfidenceError",
+    "ModelEnsemble",
+    "ModelTool",
+    "ToolRegistry",
+    "create_mcp_server",
+]
+__version__ = "0.6.3"