All scripts live in examples/. Run them with uv run python examples/<script>.py unless noted otherwise.
| Script | Purpose | Required extras |
|---|---|---|
full_pipeline_file.py |
FileBackend end-to-end: add, retrieve, expand, RetrievalOrchestrator | (none) |
full_pipeline_ob.py |
OceanBase + LangChain embedder: hybrid vector + full-text recall | oceanbase, langchain, openai |
langchain_pipeline.py |
ContextSeekMemory and ContextSeekRetriever bridge adapters |
langchain, openai |
research_agent_demo.py |
Comprehensive demo: all core features, zero external deps | (none) |
powermem_minimal.py |
Minimal PowerMem → ContextSeek plug integration | (none) |
powermem_plug_demo.py |
Full PowerMemPlug walkthrough with mixed-source retrieve() |
(none — or pip install powermem for live mode) |
llm_full_pipeline_oceanbase.py |
Real LLM calls with OceanBase: all EVOLUTION_LLM_* features end-to-end |
oceanbase, langchain, openai |
| Feature | Example(s) |
|---|---|
| InMemory / File backend | full_pipeline_file.py, research_agent_demo.py |
| OceanBase + hybrid search | full_pipeline_ob.py, llm_full_pipeline_oceanbase.py |
| LangChain bridges | langchain_pipeline.py |
| DataPlugs (RAG, memory, trace) | powermem_minimal.py, powermem_plug_demo.py |
Evolution (compact, dream, feedback) |
research_agent_demo.py, llm_full_pipeline_oceanbase.py |
| Evidence chain + provenance | research_agent_demo.py |
skill_tools() / skill_context() |
research_agent_demo.py |
| Strategy routing + canary | research_agent_demo.py |
| LLM reranking + summarization | llm_full_pipeline_oceanbase.py |
| Trace export | research_agent_demo.py |
See examples/README.md for per-script run instructions and expected output.