Metadata-Version: 2.4
Name: kestrel-cloud-gcp
Version: 0.1.0
Summary: GCP Compute Engine provider for Kestrel Sovereign — provision GPU instances, run training jobs, lifecycle management
Project-URL: Homepage, https://kestrelsovereign.com
Project-URL: Source, https://github.com/KestrelSovereignAI/kestrel-cloud-gcp
Project-URL: Issues, https://github.com/KestrelSovereignAI/kestrel-cloud-gcp/issues
Author: UncleSaurus
Maintainer: UncleSaurus
License-Expression: Apache-2.0
License-File: LICENSE
Keywords: agents,ai,compute-engine,gcp,google-cloud,gpu,kestrel,lora,training
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: Operating System :: MacOS
Classifier: Operating System :: Microsoft :: Windows
Classifier: Operating System :: POSIX :: Linux
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Classifier: Topic :: System :: Distributed Computing
Requires-Python: <3.14,>=3.11
Requires-Dist: google-cloud-compute>=1.15.0
Requires-Dist: google-cloud-logging>=3.5.0
Requires-Dist: kestrel-sovereign-sdk<1,>=0.2
Requires-Dist: kestrel-sovereign<1,>=0.5
Provides-Extra: test
Requires-Dist: pytest-asyncio>=1.1.0; extra == 'test'
Requires-Dist: pytest>=8.0.0; extra == 'test'
Description-Content-Type: text/markdown

# kestrel-cloud-gcp

GCP Compute Engine provider for Kestrel Sovereign agents. Provision GPU instances, run LoRA training jobs over SSH, and manage instance lifecycle from your agent's tools.

## Installation

```bash
uv pip install kestrel-cloud-gcp
```

The feature is auto-discovered by Kestrel Sovereign via the `kestrel_sovereign.features` entry point — install it alongside `kestrel-sovereign` and `GCPComputeFeature` registers itself at startup.

## Configuration

| Variable | Description |
|----------|-------------|
| `GCP_PROJECT_ID` | GCP project ID (required) |
| `GOOGLE_APPLICATION_CREDENTIALS` | Path to service-account JSON (or use ADC / Workload Identity) |

Optional `[gcp_compute]` section in `kestrel.toml`:

```toml
[gcp_compute.manager]
default_zone = "us-central1-a"
default_region = "us-central1"
prefer_spot = true
```

## What's provided

- `GCPComputeFeature` — agent-facing tools to provision/list/stop GPU instances, run training over SSH, monitor jobs
- Standalone API: `GCPComputeEngineManager` for direct programmatic use
- Spot-instance preference + automatic lifecycle (TTL, budget caps)

## Dependencies

- `kestrel-sovereign-sdk>=0.2,<1` — base `Feature`, `tool`, `ToolCategory`, `BackendType` interfaces
- `kestrel-sovereign>=0.5,<1` — `kestrel.toml` unified-config loader (runtime)
- `google-cloud-compute>=1.15.0`
- `google-cloud-logging>=3.5.0`

## Development

```bash
uv pip install -e '.[test]'
uv run pytest
```

## License

Apache-2.0
