Roadmap
This project is intentionally complete for its limited scope. Future work should focus on release hardening, examples and small usability improvements rather than expanding into orchestration.
Completed Milestones
- Package skeleton and typed public API.
Agentfacade.Runtimeexecution loop.- Prompt builder.
- Conversation history.
- Memory protocol and in-memory implementation.
- Tool registry and executor.
- Callback system.
- Streaming responses.
- Runtime configuration.
- Official
python-rag-frameworkintegration throughRAG.retrieve(...). - Callable and OpenAI LLM adapters.
- Examples for basic usage, tools, memory, RAG and OpenAI.
- MkDocs documentation site.
- GitHub Actions CI.
- Package build verification.
Current Status
The runtime is suitable as a small portfolio-quality Python library. It can
execute single agents, use tools, include optional memory, and consume retrieval
context from python-rag-framework.
The preferred next work is conservative:
- Keep documentation aligned with the public API.
- Add small examples based on real usage.
- Improve release notes and packaging checks.
- Fix bugs and typing issues as they appear.
Non Goals
- Multi-agent systems.
- Workflow graphs or DAG execution.
- Scheduling.
- Long-running services.
- Distributed execution.
- Cloud deployment or Kubernetes.
- Web servers, REST APIs, frontends or dashboards.