| Name | Modified | Size | Downloads / Week |
|---|---|---|---|
| Parent folder | |||
| README.md | 2026-04-11 | 1.6 kB | |
| underthesea v9.3.0 source code.tar.gz | 2026-04-11 | 59.2 MB | |
| underthesea v9.3.0 source code.zip | 2026-04-11 | 59.7 MB | |
| Totals: 3 Items | 118.9 MB | 0 | |
Highlights
Agentic AI — underthesea now ships a multi-provider agent harness with zero external LLM SDK dependencies.
Agent harness (#988)
- Multi-provider support: OpenAI, Azure OpenAI, Anthropic, Gemini — each provider is a standalone class using raw HTTP (urllib + json)
- Zero external dependencies: no
openai,anthropic, orgoogle-genaipackages required - LLM auto-detect: automatically selects provider from env vars (Azure > OpenAI > Anthropic > Gemini)
- Streaming:
Agent.stream()+chat_stream()across all providers via SSE - Session manager: multi-session agents with context reset and structured handoff
-
101 tests all passing
:::python from underthesea.agent import Agent, OpenAI, AzureOpenAI, Anthropic, Gemini, LLM
agent = Agent(name="bot", provider=AzureOpenAI(api_key=..., endpoint=..., deployment=...)) agent("Xin chào")
for chunk in agent.stream("Explain AI agents"): print(chunk, end="", flush=True)
NLP improvements
- Improve
sent_tokenizewith trained Punkt parameters (#970) - Add address converter module for post-merger admin units (#959)
Other changes
- Switch license from GPL-3.0 to Apache-2.0 (#956)
- Docs: restructure README to focus on Agent, move NLP details to NLP.md
- Docs: rewrite README in English
- Docs: add technical reports for Language Identification and POS Tagging (#954)
- Multiple dependency security bumps (#969, [#971], [#972], [#973], [#974], [#976], [#977], [#978], [#979], [#980], [#981], [#982], [#986])