Metadata-Version: 2.4
Name: dreamtime
Version: 0.1.0
Summary: Session-close memory distillation for Claude Code
Author: memory-keeper contributors
License-Expression: MIT
Project-URL: Homepage, https://github.com/2233admin/dreamtime
Project-URL: Issues, https://github.com/2233admin/dreamtime/issues
Keywords: claude-code,memory,llm,agent,dreamtime
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Classifier: Topic :: Software Development :: Libraries :: Python Modules
Requires-Python: >=3.10
Description-Content-Type: text/markdown
Requires-Dist: PyYAML>=6.0
Provides-Extra: dev
Requires-Dist: pytest; extra == "dev"
Requires-Dist: ruff; extra == "dev"

# memory-keeper

Session-close memory distillation for Claude Code.

At the end of every Claude Code session, memory-keeper reads the `.md` files you
accessed and uses an LLM to distill key insights into your persistent memory store.

## Quick start

```bash
# Install the Stop hook into ~/.claude/settings.json
uvx memory-keeper --install-hook
```

Requires [uv](https://docs.astral.sh/uv/) and an API key in `~/.claude/memory-store/config.yaml`.

## Configuration

Copy the example config and fill in your API key:

```bash
cp config.yaml.example ~/.claude/memory-store/config.yaml
```

Edit `config.yaml` — set `api.api_key` to your key.
Any OpenAI-compatible endpoint works (Anthropic, OpenAI, local relay).

## How it works

1. **Stop hook** fires `uvx memory-keeper --enqueue` at session close (async, non-blocking).
2. A background worker processes the queue and calls your LLM to distill insights.
3. Distilled memories land in `~/.claude/memory/` as Markdown files.

## License

MIT
