Metadata-Version: 2.4
Name: artifice-model-harness
Version: 0.1.0
Summary: Shared BYOM connector and schema-validated call surface for Ollama, LM Studio, and generic OpenAI-compatible endpoints.
Project-URL: Homepage, https://github.com/Muggwoffin/artifice-suite
Project-URL: Repository, https://github.com/Muggwoffin/artifice-suite
Project-URL: Issues, https://github.com/Muggwoffin/artifice-suite/issues
Author: Maurice J. Casey
License-Expression: AGPL-3.0-or-later
License-File: LICENSE
Keywords: BYOM,LLM,Ollama,OpenAI-compatible,schema validation,structured output
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Science/Research
Classifier: Operating System :: OS Independent
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Requires-Python: >=3.11
Requires-Dist: httpx>=0.27
Requires-Dist: pydantic>=2.0
Description-Content-Type: text/markdown

# model-harness

Shared connector surface for the Artifice Suite apps. Every model interaction
in `artifice-ocr`, `artifice-draft`, and `artifice-graph` must be issued
through structured, schema-validated calls rather than freeform chat — see
`CLAUDE.md` at the repo root. `artifice-transcribe` uses this package's
`ModelConnectorConfig` for its `openai`-compatible summarization calls, in
addition to its own Whisper/pyannote pipeline.

This package currently defines the shared configuration and request/response
contract (`ModelConnectorConfig`, `SchemaCall`); provider-specific transport
(Ollama, LM Studio, generic API) is implemented per-app today and is the
planned next step to consolidate here.

Install into an app's environment with:

```
pip install -e ../../packages/model-harness
```
