Metadata-Version: 2.4
Name: atp-method
Version: 0.1.0
Summary: ATP plugin: run agent-eval-case methodology cases through the platform
Author: ATP Platform Contributors
Keywords: agent-eval-case,ai-agents,atp,evaluation,methodology
Classifier: Development Status :: 3 - Alpha
Classifier: Intended Audience :: Science/Research
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.12
Requires-Dist: atp-platform>=2.1.0
Requires-Dist: pydantic>=2.0
Requires-Dist: pyyaml>=6.0
Description-Content-Type: text/markdown

# atp-method

ATP plugin that runs [`method/`](../../method/) **agent-eval-case** methodology
cases through the platform: a schema model + a loader that maps each case to an
ATP `TestDefinition`, plus (in later slices) a methodology-aware evaluator and
the format-dispatch registration so `atp test method/cases/*.yaml` just works.

See the design in [`spec/atp-method-plugin.md`](../../spec/atp-method-plugin.md).

## Status

- [x] schema model + loader (case → `TestDefinition`)
- [x] `AgentEvalCaseEvaluator` (`critical_check` then rubric)
- [x] `register()` + source dispatch + E2E (`atp test method/cases/<case>.yaml`)

## Usage

Installed as a plugin, the platform runs methodology cases directly — a single
case or a whole sweep (directory):

```bash
atp test method/cases/req-extraction --adapter=http \
  --adapter-config endpoint=http://agent:8000/execute,allow_internal=true --runs=10
```
