Metadata-Version: 2.4
Name: recall-relay
Version: 0.1.0
Summary: Bring-your-own-key generation relay and private memory client for Recall
Author: Recall contributors
License: MIT
Keywords: generative-media,gemini,backblaze,b2,genblaze,semantic-cache
Classifier: Development Status :: 3 - Alpha
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3 :: Only
Requires-Python: >=3.10
Description-Content-Type: text/markdown

# recall-relay

`recall-relay` is a small, dependency-free Python client that checks a private Recall workspace before calling Gemini image generation. On a safe reuse match it returns the stored asset; on a miss it calls Gemini with the key in your local process, then captures the resulting bytes in Recall.

## Install from GitHub

```bash
pip install "git+https://github.com/usv240/recall-generation-memory.git#subdirectory=sdk/python"
```

For local development:

```bash
pip install -e sdk/python
```

## Configure

```bash
export RECALL_URL=https://recall-production-production.up.railway.app
export RECALL_WORKSPACE_ID=ws-...
export RECALL_WORKSPACE_KEY=save-the-one-time-key
export GEMINI_API_KEY=your-google-key
```

## Use

```bash
recall-relay doctor
recall-relay gemini "Editorial cobalt glass product image" --tag launch --intent brand=Recall --intent format=1:1 --output hero.png
```

The CLI never prints credentials. `gemini` calls Google only after Recall?s private workspace returns a safe miss.

## OpenAI image generation

Set `OPENAI_API_KEY` locally and run:

```bash
recall-relay openai "Editorial cobalt glass product image" --model gpt-image-1 --output hero.png
```

## Publish to PyPI

The repository includes a secure GitHub Trusted Publishing workflow at `.github/workflows/release.yml`. Configure PyPI to trust `usv240/recall-generation-memory`, workflow `release.yml`, environment `pypi`, then publish a GitHub Release. No long-lived PyPI token is needed. Until then, the GitHub `pip install` command above installs the exact public source.
