NVIDIA-NeMo/Guardrails

📋 What's changed

• The three major features in this release are:
• Anonymous usage reporting: Basic usage reporting is documented with clear privacy boundaries and opt-out controls. The telemetry reference explains what fields are collected, what data is excluded, how local audit files work, and how to opt out with NEMO_GUARDRAILS_NO_USAGE_STATS=1, DO_NOT_TRACK=1, or the ~/.config/nemoguardrails/do_not_track file.
• LangChain decoupling: We have had many customers ask us to be less dependent on LangChain. Now, LangChain is optional. For OpenAI-compatible LLMs, we now ship with a built-in client over httpx that offers better direct support. For others, you can bring LangChain back with NEMOGUARDRAILS_LLM_FRAMEWORK=langchain, and install the matching provider package.
• IORails milestone 2: The second phase of our new, more efficient orchestration engine. This brings us closer to feature parity to LLMRails by including streaming, OpenTelemetry, and reasoning-model support. We also added _speculative generation_ in non-streaming mode, which runs input rails in parallel with the main application LLM response generation. This hides the latency of input rails while still guaranteeing response safety.
• For an exhaustive list of new features and bugfixes, please see the [release notes](https://docs.nvidia.com/nemo/guardrails/latest/about/release-notes.html) for more details.

✨ 🚀 Features

• *(iorails)* IORails support for streaming output rails ([#1765](https://github.com/NVIDIA-NeMo/Guardrails/issues/1765), [#1766](https://github.com/NVIDIA-NeMo/Guardrails/issues/1766))
• *(iorails)* IORails OpenTelemetry tracing support ([#1793](https://github.com/NVIDIA-NeMo/Guardrails/issues/1793), [#1794](https://github.com/NVIDIA-NeMo/Guardrails/issues/1794), [#1798](https://github.com/NVIDIA-NeMo/Guardrails/issues/1798))
• *(iorails)* IORails OpenTelemetry token-level metrics support ([#1812](https://github.com/NVIDIA-NeMo/Guardrails/issues/1812), [#1846](https://github.com/NVIDIA-NeMo/Guardrails/issues/1846))
• *(iorails)* IORails reasoning model support ([#1842](https://github.com/NVIDIA-NeMo/Guardrails/issues/1842), [#1843](https://github.com/NVIDIA-NeMo/Guardrails/issues/1843))
• *(llm)* Add LangChain adapter and framework registry ([#1759](https://github.com/NVIDIA-NeMo/Guardrails/issues/1759))
• *(llm)* Add streaming tool call accumulation and LLMResponse parity ([#1789](https://github.com/NVIDIA-NeMo/Guardrails/issues/1789))
• *(llm)* Add default framework with OpenAI-compatible client ([#1797](https://github.com/NVIDIA-NeMo/Guardrails/issues/1797))
• *(llm/frameworks)* Validate framework on registration ([#1863](https://github.com/NVIDIA-NeMo/Guardrails/issues/1863))
• + 6 more

🐛 🐛 Bug Fixes

• [breaking] Reject Colang 2.0 public runtime state ([#1885](https://github.com/NVIDIA-NeMo/Guardrails/issues/1885))
• *(server)* Prioritize env var API key over forwarded client header ([#1688](https://github.com/NVIDIA-NeMo/Guardrails/issues/1688))
• *(utils)* Removing extra space from UtteranceBotActionScriptUpdated ([#1708](https://github.com/NVIDIA-NeMo/Guardrails/issues/1708))
• *(actions)* Remove redundant embedding search in generate_user_intent ([#1754](https://github.com/NVIDIA-NeMo/Guardrails/issues/1754))
• *(llmrails)* Backfill embedding model params into search provider config (fixes stale KB cache) ([#1753](https://github.com/NVIDIA-NeMo/Guardrails/issues/1753))
• *(embeddings)* Persist in-memory embedding cache instance across calls ([#1755](https://github.com/NVIDIA-NeMo/Guardrails/issues/1755))
• *(ci)* Pin baseline x86-64 compiler target to prevent SIGILL on cached venvs ([#1785](https://github.com/NVIDIA-NeMo/Guardrails/issues/1785))
• *(tests)* Use asyncio.run instead of get_event_loop in middleware tests ([#1804](https://github.com/NVIDIA-NeMo/Guardrails/issues/1804))
• + 16 more

♻️ 🚜 Refactor

• *(llm)* [breaking] Atomic switch to LLMModel protocol ([#1760](https://github.com/NVIDIA-NeMo/Guardrails/issues/1760))
• *(iorails)* [breaking] Move AsyncWorkQueue from Guardrails to IORails ([#1817](https://github.com/NVIDIA-NeMo/Guardrails/issues/1817))
• *(deps)* [breaking] Demote LangChain and LangChain-providers from core to dev ([#1806](https://github.com/NVIDIA-NeMo/Guardrails/issues/1806))
• *(iorails)* Return LLMResponse(Chunk) from ModelEngine ([#1827](https://github.com/NVIDIA-NeMo/Guardrails/issues/1827))
• *(iorails)* Refactor ModelManager ([#1778](https://github.com/NVIDIA-NeMo/Guardrails/issues/1778))
• *(iorails)* Refactor Guardrails and IORails for top-level import and clean separation ([#1893](https://github.com/NVIDIA-NeMo/Guardrails/issues/1893))
• *(iorails)* Refactor RailsManager and Nemoguard Actions ([#1762](https://github.com/NVIDIA-NeMo/Guardrails/issues/1762))
• *(llm)* Rename generate/stream to generate_async/stream_async ([#1769](https://github.com/NVIDIA-NeMo/Guardrails/issues/1769))
• + 11 more

📝 📚 Documentation

• Document release notes for 0.21 and additional details ([#1726](https://github.com/NVIDIA-NeMo/Guardrails/issues/1726))
• *(middleware)* Fix incorrect example query and expected output in agent-middleware guide ([#1784](https://github.com/NVIDIA-NeMo/Guardrails/issues/1784))
• *(iorails)* OTEL Logging page ([#1807](https://github.com/NVIDIA-NeMo/Guardrails/issues/1807))
• Fix jira 407 ([#1809](https://github.com/NVIDIA-NeMo/Guardrails/issues/1809))
• Update README ([#1820](https://github.com/NVIDIA-NeMo/Guardrails/issues/1820))
• Mark LangChain integration as opt-in in 0.22 entry-point docs ([#1856](https://github.com/NVIDIA-NeMo/Guardrails/issues/1856))
• Documentation for langchain decoupling ([#1854](https://github.com/NVIDIA-NeMo/Guardrails/issues/1854))
• *(configure-rails)* Align with 0.22 DefaultFramework / LangChain split ([#1855](https://github.com/NVIDIA-NeMo/Guardrails/issues/1855))
• + 12 more

🧪 🧪 Testing

• *(llm)* Probe OpenAI API to validate _is_openai_reasoning_model ([#1814](https://github.com/NVIDIA-NeMo/Guardrails/issues/1814))
• *(llm)* Expand reasoning-model param probe + regenerate baseline ([#1838](https://github.com/NVIDIA-NeMo/Guardrails/issues/1838))
• *(telemetry)* Add smoke driver and fixtures ([#1879](https://github.com/NVIDIA-NeMo/Guardrails/issues/1879))

📦 ⚙️ Miscellaneous Tasks

• Restore original 2023-2026 copyright dates on moved files ([#1831](https://github.com/NVIDIA-NeMo/Guardrails/issues/1831))
• Include scripts in docker image ([#1902](https://github.com/NVIDIA-NeMo/Guardrails/issues/1902))

✨ 🚀 Features

• *(library)* Update Trend Micro Vision One AI Guard official endpoint ([#1546](https://github.com/NVIDIA-NeMo/Guardrails/issues/1546))
• *(llmrails)* Add check_async method for input/output rails validation ([#1605](https://github.com/NVIDIA-NeMo/Guardrails/issues/1605))
• *(server)* Make guardrails server OpenAI compatible ([#1340](https://github.com/NVIDIA-NeMo/Guardrails/issues/1340))
• *(integration)* Add GuardrailsMiddleware for LangChain agent ([#1606](https://github.com/NVIDIA-NeMo/Guardrails/issues/1606))
• *(library)* Update Fiddler Guardrails API to match new specification ([#1619](https://github.com/NVIDIA-NeMo/Guardrails/issues/1619))
• *(library)* Add CrowdStrike AIDR community integration ([#1601](https://github.com/NVIDIA-NeMo/Guardrails/issues/1601))
• *(iorails)* Introduce IORails optimized Input/Output rail engine. Supports non-streaming parallel nemoguard input/output rails (content-safety, topic-safety, jailbreak detection) ([#1638](https://github.com/NVIDIA-NeMo/Guardrails/issues/1638), [#1649](https://github.com/NVIDIA-NeMo/Guardrails/issues/1649), [#1654](https://github.com/NVIDIA-NeMo/Guardrails/issues/1654), [#1656](https://github.com/NVIDIA-NeMo/Guardrails/issues/1656), [#1658](https://github.com/NVIDIA-NeMo/Guardrails/issues/1658), [#1660](https://github.com/NVIDIA-NeMo/Guardrails/issues/1660), [#1661](https://github.com/NVIDIA-NeMo/Guardrails/issues/1661), [#1674](https://github.com/NVIDIA-NeMo/Guardrails/issues/1674))
• *(server)* Add OpenAI compatible v1/models endpoint ([#1637](https://github.com/NVIDIA-NeMo/Guardrails/issues/1637))
• + 3 more

🐛 🐛 Bug Fixes

• *(server)* Make openai an optional server-only dependency ([#1623](https://github.com/NVIDIA-NeMo/Guardrails/issues/1623))
• *(actions)* Rename generate_next_step to generate_next_steps for task-specific LLM support ([#1603](https://github.com/NVIDIA-NeMo/Guardrails/issues/1603))
• *(library)* Add `valid` alias to action results in GuardrailsAI integration ([#1578](https://github.com/NVIDIA-NeMo/Guardrails/issues/1578)) ([#1611](https://github.com/NVIDIA-NeMo/Guardrails/issues/1611))
• *(llm)* Filter stop parameter for OpenAI reasoning models ([#1653](https://github.com/NVIDIA-NeMo/Guardrails/issues/1653))
• *(logging)* Show cache hits in Stats log and fix duplicate metadata restore ([#1666](https://github.com/NVIDIA-NeMo/Guardrails/issues/1666))
• *(cache)* Make cache stats log visible in verbose mode ([#1667](https://github.com/NVIDIA-NeMo/Guardrails/issues/1667))
• *(library)* Use bot refuse to respond in gliner PII detection flows ([#1671](https://github.com/NVIDIA-NeMo/Guardrails/issues/1671))
• *(streaming)* Handle None stop tokens in streaming handler ([#1685](https://github.com/NVIDIA-NeMo/Guardrails/issues/1685))
• + 2 more

♻️ 🚜 Refactor

• *(streaming)* Remove LangChain callback dependencies from StreamingHandler ([#1547](https://github.com/NVIDIA-NeMo/Guardrails/issues/1547))
• *(streaming)* Remove ChatNVIDIA streaming patch ([#1607](https://github.com/NVIDIA-NeMo/Guardrails/issues/1607))
• *(streaming)* [breaking] Remove stream_usage and fix streaming metadata capture ([#1624](https://github.com/NVIDIA-NeMo/Guardrails/issues/1624))

⚡ ⚡ Performance

• *(actions)* Lazy initialization of embedding indexes ([#1572](https://github.com/NVIDIA-NeMo/Guardrails/issues/1572))

📦 ⚙️ Miscellaneous Tasks

• Update Pangea User-Agent repo URL ([#1595](https://github.com/NVIDIA-NeMo/Guardrails/issues/1595)) ([#1610](https://github.com/NVIDIA-NeMo/Guardrails/issues/1610))
• *(jailbreak)* Update dependencies for jailbreak detection docker container. ([#1596](https://github.com/NVIDIA-NeMo/Guardrails/issues/1596))
• Remove multi_kb example ([#1673](https://github.com/NVIDIA-NeMo/Guardrails/issues/1673))
• *(iorails)* Increase work queue concurrency and depth ([#1674](https://github.com/NVIDIA-NeMo/Guardrails/issues/1674))
• *(docs)* Remove AI Virtual Assistant Blueprint notebook ([#1682](https://github.com/NVIDIA-NeMo/Guardrails/issues/1682))
• Update dependencies ahead of v0.21 release ([#1617](https://github.com/NVIDIA-NeMo/Guardrails/issues/1617))

✨ New Contributors

• @trend-anurag-das made their first contribution in https://github.com/NVIDIA-NeMo/Guardrails/pull/1546
• @christinaexyou made their first contribution in https://github.com/NVIDIA-NeMo/Guardrails/pull/1340
• @kevinfiddlerai made their first contribution in https://github.com/NVIDIA-NeMo/Guardrails/pull/1619
• @ichbinlucaskim made their first contribution in https://github.com/NVIDIA-NeMo/Guardrails/pull/1610
• @vitchor made their first contribution in https://github.com/NVIDIA-NeMo/Guardrails/pull/1576
• Full Changelog: https://github.com/NVIDIA-NeMo/Guardrails/compare/v0.20.0...v0.21.0

✨ 🚀 Features

• *(llm)* Propagate model and base URL in LLMCallException; improve error handling ([#1502](https://github.com/NVIDIA-NeMo/Guardrails/issues/1502))
• *(content_safety)* Add support to auto select multilingual refusal bot messages ([#1530](https://github.com/NVIDIA-NeMo/Guardrails/issues/1530))
• *(library)* Adding GLiNER for PII detection (open alternative to PrivateAI) ([#1545](https://github.com/NVIDIA-NeMo/Guardrails/issues/1545))
• *(benchmark)* Implement Mock LLM streaming ([#1564](https://github.com/NVIDIA-NeMo/Guardrails/issues/1564))
• *(library)* Add reasoning guardrail connector ([#1565](https://github.com/NVIDIA-NeMo/Guardrails/issues/1565))

🐛 🐛 Bug Fixes

• *(models)* Surface relevant exception when initializing langchain model ([#1516](https://github.com/NVIDIA-NeMo/Guardrails/issues/1516))
• *(llm)* Filter temperature parameter for OpenAI reasoning models ([#1526](https://github.com/NVIDIA-NeMo/Guardrails/issues/1526))
• *(bot-thinking)* Tackle bug with reasoning trace leak across llm calls ([#1582](https://github.com/NVIDIA-NeMo/Guardrails/issues/1582))
• *(providers)* Handle langchain 1.2.1 dict type for _SUPPORTED_PROVIDERS ([#1589](https://github.com/NVIDIA-NeMo/Guardrails/issues/1589))

♻️ 🚜 Refactor

• *(streaming)* [breaking] Drop streaming field from config ([#1538](https://github.com/NVIDIA-NeMo/Guardrails/issues/1538))

📦 ⚙️ Miscellaneous Tasks

• *(test)* Reduce default pytest log level from DEBUG to WARNING ([#1523](https://github.com/NVIDIA-NeMo/Guardrails/issues/1523))
• *(docker)* Upgrade to Python 3.12-slim base image ([#1522](https://github.com/NVIDIA-NeMo/Guardrails/issues/1522))
• Run pre-commits to update license date for 2026 ([#1562](https://github.com/NVIDIA-NeMo/Guardrails/issues/1562))
• Move Benchmark code to top-level ([#1559](https://github.com/NVIDIA-NeMo/Guardrails/issues/1559))
• Update repo to <https://github.com/NVIDIA-NeMo/Guardrails> ([#1594](https://github.com/NVIDIA-NeMo/Guardrails/issues/1594))

✨ New Contributors

• @alexahaushalter made their first contribution in https://github.com/NVIDIA-NeMo/Guardrails/pull/1543
• @andreatgretel made their first contribution in https://github.com/NVIDIA-NeMo/Guardrails/pull/1545
• Full Changelog: https://github.com/NVIDIA-NeMo/Guardrails/compare/v0.19.0...v0.20.0

📋 What's Changed

• LangChain v1.x Support Is Here!
• This release also introduces several important bug fixes including an important fix related to async streaming for the `nim` or `nvidia_ai_endpoints` provider.

✨ 🚀 Features

• Support langchain v1 ([#1472](https://github.com/NVIDIA/NeMo-Guardrails/issues/1472))
• *(llm)* Add LangChain 1.x content blocks support for reasoning and tool calls ([#1496](https://github.com/NVIDIA/NeMo-Guardrails/issues/1496))
• *(benchmark)* Add Procfile to run Guardrails and mock LLMs ([#1490](https://github.com/NVIDIA/NeMo-Guardrails/issues/1490))
• *(benchmark)*: Add AIPerf run script (([#1501](https://github.com/NVIDIA/NeMo-Guardrails/issues/1501)))

🐛 🐛 Bug Fixes

• *(llm)* Add async streaming support to ChatNVIDIA provider patch ([#1504](https://github.com/NVIDIA/NeMo-Guardrails/issues/1504))
• ensure stream_async background task completes before exit ([#1508](https://github.com/NVIDIA/NeMo-Guardrails/issues/1508))
• *(cli)* Fix TypeError in v2.x chat due to incorrect State/dict conversion ([#1509](https://github.com/NVIDIA/NeMo-Guardrails/issues/1509))
• *(llmrails)*: skip output rails when dialog disabled and no bot_message provided ([#1518](https://github.com/NVIDIA/NeMo-Guardrails/issues/1518))
• *(llm)*: ensure that stop token is not ignored if llm_params is None ([#1529](https://github.com/NVIDIA/NeMo-Guardrails/issues/1529))

📦 ⚙️ Miscellaneous Tasks

• *(llm)* Remove deprecated llm_params module ([#1475](https://github.com/NVIDIA/NeMo-Guardrails/issues/1475))

📦 ◀️ Revert

• *(llm)* Remove custom HTTP headers patch now in langchain-nvidia-ai-endpoints v0.3.19 ([#1503](https://github.com/NVIDIA/NeMo-Guardrails/issues/1503))
• Full Changelog: https://github.com/NVIDIA-NeMo/Guardrails/compare/v0.18.0...v0.19.0

📋 What's Changed

• We removed support for Python 3.9, ahead of its EOL in October 2025. We invested in code quality by adding type-annotations and pre-commit checks across a majority of the codebase.

✨ 🚀 Features

• *(bot-thinking)* Implement BotThinking events to process reasoning traces in Guardrails ([#1431](https://github.com/NVIDIA/NeMo-Guardrails/issues/1431)), ([#1432](https://github.com/NVIDIA/NeMo-Guardrails/issues/1432)), ([#1434](https://github.com/NVIDIA/NeMo-Guardrails/issues/1434)).
• *(embeddings)* Add Azure OpenAI embedding provider ([#702](https://github.com/NVIDIA/NeMo-Guardrails/issues/702)).
• *(embeddings)* Add Cohere embedding integration ([#1305](https://github.com/NVIDIA/NeMo-Guardrails/issues/1305)).
• *(embeddings)* Add Google embedding integration ([#1304](https://github.com/NVIDIA/NeMo-Guardrails/issues/1304)).
• *(library)* Add Cisco AI Defense integration ([#1433](https://github.com/NVIDIA/NeMo-Guardrails/issues/1433)).
• *(cache)* Add in-memory LFU caches for content-safety, topic-control, and jailbreak detection models ([#1436](https://github.com/NVIDIA/NeMo-Guardrails/issues/1436)), ([#1456](https://github.com/NVIDIA/NeMo-Guardrails/issues/1456)), ([#1457](https://github.com/NVIDIA/NeMo-Guardrails/issues/1457)), ([#1458](https://github.com/NVIDIA/NeMo-Guardrails/issues/1458)).
• *(llm)* Add automatic provider inference for LangChain LLMs ([#1460](https://github.com/NVIDIA/NeMo-Guardrails/issues/1460)).
• *(llm)* Add custom HTTP headers support to ChatNVIDIA provider ([#1461](https://github.com/NVIDIA/NeMo-Guardrails/issues/1461)).

🐛 🐛 Bug Fixes

• *(config)* Validate content safety and topic control configs at creation time ([#1450](https://github.com/NVIDIA/NeMo-Guardrails/issues/1450)).
• *(jailbreak)* Capitalization of `Snowflake` in use of `snowflake-arctic-embed-m-long` name. ([#1464](https://github.com/NVIDIA/NeMo-Guardrails/issues/1464)).
• *(runtime)* Ensure stop flag is set for policy violations in parallel rails ([#1467](https://github.com/NVIDIA/NeMo-Guardrails/issues/1467)).
• *(llm)* [breaking] Extract reasoning traces to separate field instead of prepending ([#1468](https://github.com/NVIDIA/NeMo-Guardrails/issues/1468)).
• *(streaming)* [breaking] Raise error when stream_async used with disabled output rails streaming ([#1470](https://github.com/NVIDIA/NeMo-Guardrails/issues/1470)).
• *(llm)* Add fallback extraction for reasoning traces from <think> tags ([#1474](https://github.com/NVIDIA/NeMo-Guardrails/issues/1474)).
• *(runtime)* Set stop flag for exception-based rails in parallel mode ([#1487](https://github.com/NVIDIA/NeMo-Guardrails/issues/1487)).

♻️ 🚜 Refactor

• [breaking] Replace reasoning trace extraction with LangChain additional_kwargs ([#1427](https://github.com/NVIDIA/NeMo-Guardrails/issues/1427))

📝 📚 Documentation

• *(examples)* Add Nemoguard in-memory cache configuration example ([#1459](https://github.com/NVIDIA/NeMo-Guardrails/issues/1459)), ([#1480](https://github.com/NVIDIA/NeMo-Guardrails/issues/1480)).
• Add guide for bot reasoning guardrails ([#1479](https://github.com/NVIDIA/NeMo-Guardrails/issues/1479)).
• Update LLM reasoning traces configuration ([#1483](https://github.com/NVIDIA/NeMo-Guardrails/issues/1483)).

🧪 🧪 Testing

• Add mock embedding provider tests ([#1446](https://github.com/NVIDIA/NeMo-Guardrails/issues/1446))
• *(cli)* Add comprehensive CLI test suite and reorganize files ([#1339](https://github.com/NVIDIA/NeMo-Guardrails/issues/1339))
• Skip FastEmbed tests when not in live mode ([#1462](https://github.com/NVIDIA/NeMo-Guardrails/issues/1462))
• Fix flaky stats logging interval timing test ([#1463](https://github.com/NVIDIA/NeMo-Guardrails/issues/1463))
• Restore test that was skipped due to Colang 2.0 serialization issue ([#1449](https://github.com/NVIDIA/NeMo-Guardrails/issues/1449))

📦 ⚙️ Miscellaneous Tasks

• Resolve PyPI publish workflow trigger and reliability issues ([#1443](https://github.com/NVIDIA/NeMo-Guardrails/issues/1443))
• Fix sparse checkout for publish pypi workflow ([#1444](https://github.com/NVIDIA/NeMo-Guardrails/issues/1444))
• Drop Python 3.9 support ahead of October 2025 EOL ([#1426](https://github.com/NVIDIA/NeMo-Guardrails/issues/1426))
• *(types)* Add type-annotations and pre-commit checks for tracing ([#1388](https://github.com/NVIDIA/NeMo-Guardrails/issues/1388)), logging ([#1395](https://github.com/NVIDIA/NeMo-Guardrails/issues/1395)), kb ([#1385](https://github.com/NVIDIA/NeMo-Guardrails/issues/1385)), cli ([#1380](https://github.com/NVIDIA/NeMo-Guardrails/issues/1380)), embeddings ([#1383](https://github.com/NVIDIA/NeMo-Guardrails/issues/1383)), server ([#1397](https://github.com/NVIDIA/NeMo-Guardrails/issues/1397)), and llm ([#1394](https://github.com/NVIDIA/NeMo-Guardrails/issues/1394)) code.
• Update insert licenser pe-commit-hooks to use current year ([#1452](https://github.com/NVIDIA/NeMo-Guardrails/issues/1452)).
• *(library)* Remove unused vllm requirements.txt files ([#1466](https://github.com/NVIDIA/NeMo-Guardrails/issues/1466)).

✨ New Contributors

• @shubhiroy made their first contribution in https://github.com/NVIDIA-NeMo/Guardrails/pull/702
• @rucpande made their first contribution in https://github.com/NVIDIA-NeMo/Guardrails/pull/1433
• @hazai made their first contribution in https://github.com/NVIDIA-NeMo/Guardrails/pull/1436
• Full Changelog: https://github.com/NVIDIA-NeMo/Guardrails/compare/v0.17.0...v0.18.0

📋 What's Changed

• We also enhanced our support for integrating with LangChain RunnableRails. This release supports synchronous and asynchronous operations, streaming, and batch processing.

✨ 🚀 Features

• *(tool-calling)* Add tool call passthrough support in LLMRails ([#1364](https://github.com/NVIDIA/NeMo-Guardrails/issues/1364))
• *(runnable-rails)* Complete rewrite of RunnableRails with full LangChain Runnable protocol support ([#1366](https://github.com/NVIDIA/NeMo-Guardrails/issues/1366), [#1369](https://github.com/NVIDIA/NeMo-Guardrails/issues/1369), [#1370](https://github.com/NVIDIA/NeMo-Guardrails/issues/1370), [#1405](https://github.com/NVIDIA/NeMo-Guardrails/issues/1405))
• *(tool-rails)* Add support for tool output rails and validation ([#1382](https://github.com/NVIDIA/NeMo-Guardrails/issues/1382))
• *(tool-rails)* Implement tool input rails for tool message validation and processing ([#1386](https://github.com/NVIDIA/NeMo-Guardrails/issues/1386))
• *(library)* Add Trend Micro Vision One AI Application Security community integration ([#1355](https://github.com/NVIDIA/NeMo-Guardrails/issues/1355))
• *(llm)* Pass llm params directly ([#1387](https://github.com/NVIDIA/NeMo-Guardrails/issues/1387))

🐛 🐛 Bug Fixes

• *(jailbreak)* Handle URL joining with/without trailing slashes ([#1346](https://github.com/NVIDIA/NeMo-Guardrails/issues/1346))
• *(logging)* Handle missing id and task in verbose logs ([#1343](https://github.com/NVIDIA/NeMo-Guardrails/issues/1343))
• *(library)* Fix import package declaration to new cleanlab-tlm name ([#1401](https://github.com/NVIDIA/NeMo-Guardrails/issues/1401))
• *(logging)* Add "Tool" type to message sender labeling ([#1412](https://github.com/NVIDIA/NeMo-Guardrails/issues/1412))
• *(logging)* Correct message type formatting in logs ([#1416](https://github.com/NVIDIA/NeMo-Guardrails/issues/1416))

♻️ 🚜 Refactor

• *(llm)* Remove LLMs isolation for actions ([#1408](https://github.com/NVIDIA/NeMo-Guardrails/issues/1408))

📝 📚 Documentation

• *(examples)* Add NeMoGuard safety rails config example for Colang 1.0 ([#1365](https://github.com/NVIDIA/NeMo-Guardrails/issues/1365))
• Add hardware reqs ([#1411](https://github.com/NVIDIA/NeMo-Guardrails/issues/1411))
• Add tools integration guide ([#1414](https://github.com/NVIDIA/NeMo-Guardrails/issues/1414))
• *(langgraph)* Add integration guide for LangGraph ([#1422](https://github.com/NVIDIA/NeMo-Guardrails/issues/1422))
• *(langchain)* Update with full support and add tool calling guide … ([#1419](https://github.com/NVIDIA/NeMo-Guardrails/issues/1419))
• *(langgraph)* Clarify tool examples and replace calculate_math with multiply ([#1439](https://github.com/NVIDIA/NeMo-Guardrails/issues/1439))

📦 ⚙️ Miscellaneous Tasks

• *(docs)* Update v0.16.0 release date in changelog ([#1377](https://github.com/NVIDIA/NeMo-Guardrails/issues/1377))
• *(docs)* Add link to demo.py script in Getting-Started section ([#1399](https://github.com/NVIDIA/NeMo-Guardrails/issues/1399))
• *(types)* Type-clean rails (86 errors) ([#1396](https://github.com/NVIDIA/NeMo-Guardrails/issues/1396))
• *(jailbreak-detection)* Update transformers and torch ([#1417](https://github.com/NVIDIA/NeMo-Guardrails/issues/1417))
• *(types)* Type-clean /actions (189 errors) ([#1361](https://github.com/NVIDIA/NeMo-Guardrails/issues/1361))
• *(docs)* Update repository owner ([#1425](https://github.com/NVIDIA/NeMo-Guardrails/issues/1425))

✨ Key Features

• Enhanced tracing system with [OpenTelemetry semantic conventions](https://opentelemetry.io/docs/specs/semconv/gen-ai/). To learn more, refer to [](tracing). For usage examples, refer to the following notebooks
• [Tracing Guardrails Quickstart](https://github.com/NVIDIA/NeMo-Guardrails/tree/develop/docs/getting-started/8-tracing/1_tracing_quickstart.ipynb)
• [Tracing Guardrails with Jaeger](https://github.com/NVIDIA/NeMo-Guardrails/tree/develop/docs/getting-started/8-tracing/2_tracing_with_jaeger.ipynb)
• Community integration with [GuardrailsAI](https://www.guardrailsai.com/) and [Pangea AI Guard](https://pangea.cloud/services/ai-guard).

📋 Other Changes

• Added documentation about using KV cache reuse for LLM-based NemoGuard NIMs. By using KV cache reuse, you can improve the performance of LLM-based NemoGuard NIMs where the system prompt is the same for all calls up to the point where user query and LLM response are injected. To learn more, refer to [](kv-cache-reuse).
• ----------

✨ 🚀 Features

• *(llmrails)* Support method chaining by returning self from LLMRails.register_* methods ([#1296](https://github.com/NVIDIA/NeMo-Guardrails/issues/1296))
• Add Pangea AI Guard community integration ([#1300](https://github.com/NVIDIA/NeMo-Guardrails/issues/1300))
• *(llmrails)* Isolate LLMs only for configured actions ([#1342](https://github.com/NVIDIA/NeMo-Guardrails/issues/1342))
• Enhance tracing system with OpenTelemetry semantic conventions ([#1331](https://github.com/NVIDIA/NeMo-Guardrails/issues/1331))
• Add GuardrailsAI community integration ([#1298](https://github.com/NVIDIA/NeMo-Guardrails/issues/1298))

🐛 🐛 Bug Fixes

• *(models)* Suppress langchain_nvidia_ai_endpoints warnings ([#1371](https://github.com/NVIDIA/NeMo-Guardrails/issues/1371))
• *(tracing)* Respect the user-provided log options regardless of tracing configuration
• *(config)* Ensure adding RailsConfig objects handles None values ([#1328](https://github.com/NVIDIA/NeMo-Guardrails/issues/1328))
• *(config)* Add handling for config directory with `.yml`/`.yaml` extension ([#1293](https://github.com/NVIDIA/NeMo-Guardrails/issues/1293))
• *(colang)* Apply guardrails transformations to LLM inputs and bot outputs. ([#1297](https://github.com/NVIDIA/NeMo-Guardrails/issues/1297))
• *(topic_safety)* Handle InternalEvent objects in topic safety actions for Colang 2.0 ([#1335](https://github.com/NVIDIA/NeMo-Guardrails/issues/1335))
• *(prompts)* Prevent IndexError when LLM provided via constructor with empty models config ([#1334](https://github.com/NVIDIA/NeMo-Guardrails/issues/1334))
• *(llmrails)* Handle LLM models without model_kwargs field in isolation ([#1336](https://github.com/NVIDIA/NeMo-Guardrails/issues/1336))
• + 1 more

♻️ 🚜 Refactor

• *(llm)* Move get_action_details_from_flow_id from llmrails.py to utils.py ([#1341](https://github.com/NVIDIA/NeMo-Guardrails/issues/1341))

📝 📚 Documentation

• Integrate with multilingual NIM ([#1354](https://github.com/NVIDIA/NeMo-Guardrails/issues/1354))
• *(tracing)* Update tracing notebooks with VDR feedback ([#1376](https://github.com/NVIDIA/NeMo-Guardrails/issues/1376))
• Add kv cache reuse documentation ([#1330](https://github.com/NVIDIA/NeMo-Guardrails/issues/1330))
• *(examples)* Add Colang 2.0 example for sensitive data detection ([#1301](https://github.com/NVIDIA/NeMo-Guardrails/issues/1301))
• Add extra slash to jailbreak detect nim_base_url([#1345](https://github.com/NVIDIA/NeMo-Guardrails/issues/1345))
• Add tracing notebook ([#1337](https://github.com/NVIDIA/NeMo-Guardrails/issues/1337))
• Jaeger tracing notebook ([#1353](https://github.com/NVIDIA/NeMo-Guardrails/issues/1353))
• *(examples)* Add NeMoGuard rails config for colang 2 ([#1289](https://github.com/NVIDIA/NeMo-Guardrails/issues/1289))
• + 2 more

🧪 🧪 Testing

• *(jailbreak)* Add missing pytest.mark.asyncio decorators ([#1352](https://github.com/NVIDIA/NeMo-Guardrails/issues/1352))

📦 ⚙️ Miscellaneous Tasks

• *(docs)* Rename test_csl.py to csl.py ([#1347](https://github.com/NVIDIA/NeMo-Guardrails/issues/1347))

✨ 🚀 Features

• *(tracing)* [breaking] Update tracing to use otel api ([#1269](https://github.com/NVIDIA/NeMo-Guardrails/issues/1269))
• *(streaming)* Implement parallel streaming output rails execution ([#1263](https://github.com/NVIDIA/NeMo-Guardrails/issues/1263), [#1324](https://github.com/NVIDIA/NeMo-Guardrails/pull/1324))
• *(streaming)* Support external async token generators ([#1286](https://github.com/NVIDIA/NeMo-Guardrails/issues/1286))
• Support parallel rails execution ([#1234](https://github.com/NVIDIA/NeMo-Guardrails/issues/1234), [#1323](https://github.com/NVIDIA/NeMo-Guardrails/pull/1323))

🐛 🐛 Bug Fixes

• *(streaming)* Resolve word concatenation in streaming output rails ([#1259](https://github.com/NVIDIA/NeMo-Guardrails/issues/1259))
• *(streaming)* Enable token usage tracking for streaming LLM calls ([#1264](https://github.com/NVIDIA/NeMo-Guardrails/issues/1264), [#1285](https://github.com/NVIDIA/NeMo-Guardrails/issues/1285))
• *(tracing)* Prevent mutation of user options when tracing is enabled ([#1273](https://github.com/NVIDIA/NeMo-Guardrails/issues/1273))
• *(rails)* Prevent LLM parameter contamination in rails ([#1306](https://github.com/NVIDIA/NeMo-Guardrails/issues/1306))

📝 📚 Documentation

• Release notes 0.14.1 ([#1272](https://github.com/NVIDIA/NeMo-Guardrails/issues/1272))
• Update guardrails-library.md to include Clavata as a third party API ([#1294](https://github.com/NVIDIA/NeMo-Guardrails/issues/1294))
• *(streaming)* Add section on token usage tracking ([#1282](https://github.com/NVIDIA/NeMo-Guardrails/issues/1282))
• Add parallel rail section and split config page ([#1295](https://github.com/NVIDIA/NeMo-Guardrails/issues/1295))
• Show complete prompts.yml content in getting started tutorial ([#1311](https://github.com/NVIDIA/NeMo-Guardrails/issues/1311))
• *(tracing)* Update and streamline tracing guide ([#1307](https://github.com/NVIDIA/NeMo-Guardrails/issues/1307))

📦 ⚙️ Miscellaneous Tasks

• *(dependabot)* Remove dependabot configuration ([#1281](https://github.com/NVIDIA/NeMo-Guardrails/issues/1281))
• *(CI)* Add release workflow ([#1309](https://github.com/NVIDIA/NeMo-Guardrails/issues/1309), [#1318](https://github.com/NVIDIA/NeMo-Guardrails/issues/1318))

✨ New Contributors

• @ashwini made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/1294
• Full Changelog: https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.14.1...v0.15.0

📋 What's Changed

• This patch release addresses several important bug fixes and improvements, particularly focusing on jailbreak detection, LLMRails configuration handling, and content safety compatibility issues.

✨ 🚀 Features

• *(jailbreak)* Add direct API key configuration support ([#1260](https://github.com/NVIDIA/NeMo-Guardrails/issues/1260))

🐛 🐛 Bug Fixes

• *(jailbreak)* Lazy load jailbreak detection dependencies ([#1223](https://github.com/NVIDIA/NeMo-Guardrails/issues/1223))
• *(llmrails)* Constructor LLM should not skip loading other config models ([#1221](https://github.com/NVIDIA/NeMo-Guardrails/issues/1221), [#1247](https://github.com/NVIDIA/NeMo-Guardrails/issues/1247), [#1250](https://github.com/NVIDIA/NeMo-Guardrails/issues/1250), [#1258](https://github.com/NVIDIA/NeMo-Guardrails/issues/1258))
• *(content_safety)* Replace try-except with iterable unpacking for policy violations ([#1207](https://github.com/NVIDIA/NeMo-Guardrails/issues/1207))
• *(jailbreak)* Pin numpy==1.23.5 for scikit-learn compatibility ([#1249](https://github.com/NVIDIA/NeMo-Guardrails/issues/1249))
• *(output_parsers)* Iterable unpacking compatibility in content safety parsers ([#1242](https://github.com/NVIDIA/NeMo-Guardrails/issues/1242))

📝 📚 Documentation

• More heading levels so RNs resolve links ([#1228](https://github.com/NVIDIA/NeMo-Guardrails/issues/1228))
• Update docs version ([#1219](https://github.com/NVIDIA/NeMo-Guardrails/issues/1219))
• Fix jailbreak detection build instructions ([#1248](https://github.com/NVIDIA/NeMo-Guardrails/issues/1248))
• Change ABC bot link at docs ([#1261]([#1248](https://github.com/NVIDIA/NeMo-Guardrails/issues/1261)))

🧪 🧪 Testing

• Fix async test failures in cache embeddings and buffer strategy tests ([#1237](https://github.com/NVIDIA/NeMo-Guardrails/issues/1237))
• *(content_safety)* Add tests for content safety actions ([#1240](https://github.com/NVIDIA/NeMo-Guardrails/issues/1240))

📦 ⚙️ Miscellaneous Tasks

• Update pre-commit-hooks to v5.0.0 ([#1238](https://github.com/NVIDIA/NeMo-Guardrails/issues/1238))

✨ New Contributors

• @jeffreyscarpenter made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/1223
• @gcr made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/1207
• @bwook00 made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/1261
• Full Changelog: https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.14.0...v0.14.1

📋 What's Changed

• From the community, a new 3rd-party guardrail connects to [Clavata](https://www.clavata.ai/)’s customized LLM content moderation.
• Among other improvements and bug fixes, this release also now supports Python 3.13.

✨ 🚀 Features

• Change topic following prompt to allow chitchat ([#1097](https://github.com/NVIDIA/NeMo-Guardrails/issues/1097))
• Validate model name configuration ([#1084](https://github.com/NVIDIA/NeMo-Guardrails/issues/1084))
• Add support for langchain partner and community chat models ([#1085](https://github.com/NVIDIA/NeMo-Guardrails/issues/1085))
• Add fuzzy find provider capability to cli ([#1088](https://github.com/NVIDIA/NeMo-Guardrails/issues/1088))
• Add code injection detection to guardrails library ([#1091](https://github.com/NVIDIA/NeMo-Guardrails/issues/1091))
• Add clavata community integration ([#1027](https://github.com/NVIDIA/NeMo-Guardrails/issues/1027))
• Implement validation to forbid dialog rails with reasoning traces ([#1137](https://github.com/NVIDIA/NeMo-Guardrails/issues/1137))
• Load yara lazily to avoid action dispatcher error ([#1162](https://github.com/NVIDIA/NeMo-Guardrails/issues/1162))
• + 7 more

🐛 🐛 Bug Fixes

• Correct task name for self_check_facts ([#1040](https://github.com/NVIDIA/NeMo-Guardrails/issues/1040))
• Error in LLMRails with tracing enabled ([#1103](https://github.com/NVIDIA/NeMo-Guardrails/issues/1103))
• Self check output colang 1 flow ([#1126](https://github.com/NVIDIA/NeMo-Guardrails/issues/1126))
• Use ValueError in TaskPrompt to resolve TypeError raised by Pydantic ([#1132](https://github.com/NVIDIA/NeMo-Guardrails/issues/1132))
• Correct dialog rails activation logic ([#1161](https://github.com/NVIDIA/NeMo-Guardrails/issues/1161))
• Allow reasoning traces when embeddings_only is True ([#1170](https://github.com/NVIDIA/NeMo-Guardrails/issues/1170))
• Prevent explain_info overwrite during stream_async ([#1194](https://github.com/NVIDIA/NeMo-Guardrails/issues/1194))
• Colang 2 issues in community integrations ([#1140](https://github.com/NVIDIA/NeMo-Guardrails/issues/1140))
• + 1 more

♻️ 🚜 Refactor

• Reorganize HuggingFace provider structure ([#1083](https://github.com/NVIDIA/NeMo-Guardrails/issues/1083))
• Remove support for deprecated nemollm engine ([#1076](https://github.com/NVIDIA/NeMo-Guardrails/issues/1076))
• [breaking] Remove deprecated return_context argument ([#1147](https://github.com/NVIDIA/NeMo-Guardrails/issues/1147))
• Rename `remove_thinking_traces` field to `remove_reasoning_traces` ([#1176](https://github.com/NVIDIA/NeMo-Guardrails/issues/1176))
• Update deprecated field handling for remove_thinking_traces ([#1196](https://github.com/NVIDIA/NeMo-Guardrails/issues/1196))
• Introduce END_OF_STREAM sentinel and update handling ([#1185](https://github.com/NVIDIA/NeMo-Guardrails/issues/1185))

📝 📚 Documentation

• Remove markup from code block ([#1081](https://github.com/NVIDIA/NeMo-Guardrails/issues/1081))
• Replace img tag with Markdown images ([#1087](https://github.com/NVIDIA/NeMo-Guardrails/issues/1087))
• Remove NeMo Service (nemollm) documentation ([#1077](https://github.com/NVIDIA/NeMo-Guardrails/issues/1077))
• Update cleanlab integration description ([#1080](https://github.com/NVIDIA/NeMo-Guardrails/issues/1080))
• Add providers fuzzy search cli command ([#1089](https://github.com/NVIDIA/NeMo-Guardrails/issues/1089))
• Clarify purpose of model parameters field in configuration guide ([#1181](https://github.com/NVIDIA/NeMo-Guardrails/issues/1181))
• Output rails are supported with streaming ([#1007](https://github.com/NVIDIA/NeMo-Guardrails/issues/1007))
• Add mention of Nemotron ([#1200](https://github.com/NVIDIA/NeMo-Guardrails/issues/1200))
• + 3 more

📦 ⚙️ Miscellaneous Tasks

• Dynamically set version using importlib.metadata ([#1072](https://github.com/NVIDIA/NeMo-Guardrails/issues/1072))
• Add link to topic control config and prompts ([#1098](https://github.com/NVIDIA/NeMo-Guardrails/issues/1098))
• Reorganize GitHub workflows for better test coverage ([#1079](https://github.com/NVIDIA/NeMo-Guardrails/issues/1079))
• Add summary jobs for workflow branch protection ([#1120](https://github.com/NVIDIA/NeMo-Guardrails/issues/1120))
• Add Adobe Analytics configuration ([#1138](https://github.com/NVIDIA/NeMo-Guardrails/issues/1138))
• Fix and revert poetry lock to its stable state ([#1133](https://github.com/NVIDIA/NeMo-Guardrails/issues/1133))
• Add Codecov integration to workflows ([#1143](https://github.com/NVIDIA/NeMo-Guardrails/issues/1143))
• Add Python 3.12 and 3.13 test jobs to gitlab workflow ([#1171](https://github.com/NVIDIA/NeMo-Guardrails/issues/1171))
• + 2 more

✨ New Contributors

• @thisthat made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/1103
• @jwmueller made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/1080
• @ilias-t made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/1027
• @smruthi33 made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/1106
• @andompesta made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/1194
• Full Changelog: https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.13.0...v0.14.0

✨ 🚀 Features

• Support models with reasoning traces ([#996](https://github.com/NVIDIA/NeMo-Guardrails/issues/996)) by @trebedea
• Add SHA-256 hashing option ([#988](https://github.com/NVIDIA/NeMo-Guardrails/issues/988)) by @mdambski
• Add Fiddler Guardrails integration ([#964](https://github.com/NVIDIA/NeMo-Guardrails/issues/964), [#1043](https://github.com/NVIDIA/NeMo-Guardrails/issues/1043)) by @copperstick6
• Add generation metadata to streaming chunks ([#1011](https://github.com/NVIDIA/NeMo-Guardrails/issues/1011)) by @Pouyanpi
• Improve alpha to beta bot migration ([#878](https://github.com/NVIDIA/NeMo-Guardrails/issues/878)) by @schuellc-nvidia
• Support multimodal input and output rails ([#1033](https://github.com/NVIDIA/NeMo-Guardrails/issues/1033)) by @Pouyanpi
• Add support for NemoGuard JailbreakDetect NIM. ([#1038](https://github.com/NVIDIA/NeMo-Guardrails/issues/1038)) by @erickgalinkin
• Set default start and end reasoning tokens ([#1050](https://github.com/NVIDIA/NeMo-Guardrails/issues/1050)) by @Pouyanpi
• + 1 more

🐛 🐛 Bug Fixes

• Ensure parse_task_output is called after all llm_call invocations ([#1047](https://github.com/NVIDIA/NeMo-Guardrails/issues/1047)) by @Pouyanpi
• Handle exceptions in generate_events to propagate errors in streaming ([#1012](https://github.com/NVIDIA/NeMo-Guardrails/issues/1012)) by @Pouyanpi
• Ensure output rails streaming is enabled explicitly ([#1045](https://github.com/NVIDIA/NeMo-Guardrails/issues/1045)) by @Pouyanpi
• Improve multimodal prompt length calculation for base64 images ([#1053](https://github.com/NVIDIA/NeMo-Guardrails/issues/1053)) by @Pouyanpi

♻️ 🚜 Refactor

• Move startup and shutdown logic to lifespan in server ([#999](https://github.com/NVIDIA/NeMo-Guardrails/issues/999)) by @xiaobo8204

📝 📚 Documentation

• Add multimodal rails documentation ([#1061](https://github.com/NVIDIA/NeMo-Guardrails/issues/1061)) by @mikemckiernan
• Add content safety tutorial ([#1042](https://github.com/NVIDIA/NeMo-Guardrails/issues/1042)) by @tgasser-nv
• Revise reasoning model info ([#1062](https://github.com/NVIDIA/NeMo-Guardrails/issues/1062)) by @mikemckiernan
• Consider new GS experience ([#1005](https://github.com/NVIDIA/NeMo-Guardrails/issues/1005)) by @mikemckiernan
• Restore deleted configuration files ([#963](https://github.com/NVIDIA/NeMo-Guardrails/issues/963)) by @Pouyanpi

📦 ⚙️ Miscellaneous Tasks

• Add Python 3.12 support ([#984](https://github.com/NVIDIA/NeMo-Guardrails/issues/984)) by @Pouyanpi

✨ New Contributors

• @mdambski made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/988
• @copperstick6 made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/964
• @xiaobo8204 made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/999
• @tgasser-nv made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/1042
• Full Changelog: https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.12.0...v0.13.0

✨ 🚀 Features

• Support Output Rails Streaming ([#966](https://github.com/NVIDIA/NeMo-Guardrails/issues/966), [#1003](https://github.com/NVIDIA/NeMo-Guardrails/issues/1003)) by @Pouyanpi
• Add unified output mapping for actions ([#965](https://github.com/NVIDIA/NeMo-Guardrails/issues/965)) by @Pouyanpi
• Add output rails support to activefence integration ([#940](https://github.com/NVIDIA/NeMo-Guardrails/issues/940)) by @noamlevy81
• Add Prompt Security integration ([#920](https://github.com/NVIDIA/NeMo-Guardrails/issues/920)) by @lior-ps
• Add pii masking capability to PrivateAI integration ([#901](https://github.com/NVIDIA/NeMo-Guardrails/issues/901)) by @letmerecall
• Add embedding_params to BasicEmbeddingsIndex ([#898](https://github.com/NVIDIA/NeMo-Guardrails/issues/898)) by @Pouyanpi
• Add score threshold to AnalyzerEngine ([#845](https://github.com/NVIDIA/NeMo-Guardrails/issues/845)) by @Pouyanpi

🐛 🐛 Bug Fixes

• Fix dependency resolution issues in AlignScore Dockerfile([#1002](https://github.com/NVIDIA/NeMo-Guardrails/issues/1002), [#982](https://github.com/NVIDIA/NeMo-Guardrails/issues/982)) by @Pouyanpi
• Fix JailbreakDetect docker files([#981](https://github.com/NVIDIA/NeMo-Guardrails/issues/981), [#1001](https://github.com/NVIDIA/NeMo-Guardrails/pull/1001)) by @erickgalinkin
• Fix TypeError from attempting to unpack already-unpacked dictionary ([#959](https://github.com/NVIDIA/NeMo-Guardrails/issues/959)) by @erickgalinkin
• Fix token stats usage in LLM call info ([#953](https://github.com/NVIDIA/NeMo-Guardrails/issues/953)) by @trebedea
• Handle unescaped quotes in generate_value using safe_eval ([#946](https://github.com/NVIDIA/NeMo-Guardrails/issues/946)) by @milk333445
• Handle non-relative file paths ([#897](https://github.co/NVIDIA/NeMo-Guardrails/issues/897)) by @Pouyanpi

📝 📚 Documentation

• Output streaming ([#976](https://github.com/NVIDIA/NeMo-Guardrails/issues/976)) by @mikemckiernan
• Fix typos with oauthtoken ([#957](https://github.com/NVIDIA/NeMo-Guardrails/issues/957)) by @Pouyanpi
• Fix broken link in prompt security ([#978](https://github.com/NVIDIA/NeMo-Guardrails/issues/978)) by @lior-ps
• Update advanced user guides per v0.11.1 doc release ([#937](https://github.com/NVIDIA/NeMo-Guardrails/issues/937)) by @Pouyanpi

📦 ⚙️ Miscellaneous Tasks

• Tolerate prompt in code blocks ([#1004](https://github.com/NVIDIA/NeMo-Guardrails/issues/1004)) by @mikemckiernan
• Update YAML indent to use two spaces ([#1009](https://github.com/NVIDIA/NeMo-Guardrails/issues/1009)) by @mikemckiernan

✨ New Contributors

• @milk333445 made their first contribution in [#946](https://github.com/NVIDIA/NeMo-Guardrails/pull/946)
• @lior-ps made their first contribution in [#920](https://github.com/NVIDIA/NeMo-Guardrails/pull/920)
• Full Changelog: <https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.11.1...v0.12.0>

📋 Changes

• [Content safety NIM microservice](https://build.nvidia.com/nvidia/llama-3_1-nemoguard-8b-content-safety) that safeguards AI against generating biased or harmful outputs, ensuring responses align with ethical standards.
• [Topic control NIM microservice](https://build.nvidia.com/nvidia/llama-3_1-nemoguard-8b-topic-control) that keeps conversations focused on approved topics, avoiding digression or inappropriate content.
• [Jailbreak detection NIM microservice](https://build.nvidia.com/nvidia/nemoguard-jailbreak-detect) that adds protection against jailbreak attempts, helping maintain AI integrity in adversarial scenarios.

✨ Added

• ContentSafety: Add ContentSafety NIM connector ([#930](https://github.com/NVIDIA/NeMo-Guardrails/pull/930)) by @prasoonvarshney
• TopicControl: Add TopicControl NIM connector ([#930](https://github.com/NVIDIA/NeMo-Guardrails/pull/930)) by @makeshn
• JailbreakDetect: Add jailbreak detection NIM connector ([#930](https://github.com/NVIDIA/NeMo-Guardrails/pull/930)) by @erickgalinkin

📋 Changed

• AutoAlign Integration: Add further enhancements and refactoring to AutoAlign integration ([#867](https://github.com/NVIDIA/NeMo-Guardrails/pull/867)) by @KimiJL

🐛 Fixed

• PrivateAI Integration: Fix Incomplete URL substring sanitization Error ([#883](https://github.com/NVIDIA/NeMo-Guardrails/pull/883)) by @NJ-186
• Content Safety: Fix typo in content safety prompts ([#873](https://github.com/NVIDIA/NeMo-Guardrails/pull/873)) by @krishna-sreeraj-tw

📝 Documentation

• NVIDIA Blueprint: Add Safeguarding AI Virtual Assistant NIM Blueprint NemoGuard NIMs ([#932](https://github.com/NVIDIA/NeMo-Guardrails/pull/932)) by @abodhankar
• ActiveFence Integration: Fix flow definition in community docs ([#890](https://github.com/NVIDIA/NeMo-Guardrails/pull/890)) by @noamlevy81

✨ New Contributors

• @krishna-sreeraj-tw made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/873
• @KimiJL made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/867
• @NJ-186 made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/883
• @noamlevy81 made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/890
• Full Changelog: https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.11.0...v0.11.1

✨ Added

• Observability: Add observability support with support for different backends ([#844](https://github.com/NVIDIA/NeMo-Guardrails/pull/844)) by @Pouyanpi
• Private AI Integration: Add Private AI Integration ([#815](https://github.com/NVIDIA/NeMo-Guardrails/pull/815)) by @letmerecall
• Patronus Evaluate API Integration: Patronus Evaluate API Integration ([#834](https://github.com/NVIDIA/NeMo-Guardrails/pull/834)) by @varjoshi
• railsignore: Add support for .railsignore file ([#790](https://github.com/NVIDIA/NeMo-Guardrails/pull/790)) by @ajanitshimanga
• Passthrough Support: Add passthrough support in Colang 2 (#779) by @Pouyanpi

📋 Changed

• Sandboxed Environment in Jinja2: Add sandboxed environment in Jinja2 ([#799](https://github.com/NVIDIA/NeMo-Guardrails/pull/799)) by @Pouyanpi
• Langchain 3 support: Upgrade LangChain to Version 0.3 ([#784](https://github.com/NVIDIA/NeMo-Guardrails/pull/784)) by @Pouyanpi
• Python 3.8: Drop support for Python 3.8 ([#803](https://github.com/NVIDIA/NeMo-Guardrails/pull/803)) by @Pouyanpi
• vllm: Bump vllm from 0.2.7 to 0.5.5 for llama_guard and patronusai([#836](https://github.com/NVIDIA/NeMo-Guardrails/pull/836))
• Colang 2.0-beta.5: Upgrade Colang 2 version including new attention library besides other features and bugfixes ([Colang 2 Changelog](https://github.com/NVIDIA/NeMo-Guardrails/blob/develop/CHANGELOG-Colang.md))

🐛 Fixed

• Colang: Multiline string values in interaction history prompting ([#765](https://github.com/NVIDIA/NeMo-Guardrails/pull/765)) by @radinshayanfar
• Guardrails Library documentation": Fix a typo in guardrails library documentation ([#793](https://github.com/NVIDIA/NeMo-Guardrails/pull/793)) by @vedantnaik19
• Contributing Guide: Fix incorrect folder name & pre-commit setup in CONTRIBUTING.md ([#800](https://github.com/NVIDIA/NeMo-Guardrails/pull/800)) by @SSK-14
• Contributing Guide: Added correct Python command version in documentation([#801](https://github.com/NVIDIA/NeMo-Guardrails/pull/801)) by @ravinder-tw
• retrieve chunk action: Fix presence of new line in retrieve chunk action ([#809](https://github.com/NVIDIA/NeMo-Guardrails/pull/809)) by @Pouyanpi
• Standard Library import: Fix guardrails standard library import path in Colang 2.0 ([#835](https://github.com/NVIDIA/NeMo-Guardrails/pull/835)) by @Pouyanpi
• AlignScore Dockerfile: Add nltk's punkt_tab in align_score Dockerfile ([#841](https://github.com/NVIDIA/NeMo-Guardrails/pull/841)) by @yonromai
• Eval dependencies: Make pandas version constraint explicit for eval optional dependency ([#847](https://github.com/NVIDIA/NeMo-Guardrails/pull/847)) by @Pouyanpi
• + 4 more

📝 Documentation

• Installation Guide: Update optional dependencies install ([#853](https://github.com/NVIDIA/NeMo-Guardrails/pull/853)) by @Pouyanpi
• Documentation Restructuring: Restructure the docs and several style enhancements ([#855](https://github.com/NVIDIA/NeMo-Guardrails/pull/855)) by @Pouyanpi
• Got It AI deprecation: Add deprecation notice for Got It AI integration ([#857](https://github.com/NVIDIA/NeMo-Guardrails/pull/857)) by @mlmonk

✨ New Contributors

• @ravinder-tw made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/801
• @SSK-14 made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/800
• @ajanitshimanga made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/790
• @letmerecall made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/815
• @emmanuel-ferdman made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/751
• @yonromai made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/841
• Full Changelog: https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.10.1...v0.11.0

📋 What's Changed

• Fix/colang 2/flow continuation prompt by @schuellc-nvidia in https://github.com/NVIDIA/NeMo-Guardrails/pull/775
• Remove the user intent from llm library module by @schuellc-nvidia in https://github.com/NVIDIA/NeMo-Guardrails/pull/777
• Full Changelog: https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.10.0...v0.10.1

📋 What's Changed

• bug fix by @saradiazdelser in https://github.com/NVIDIA/NeMo-Guardrails/pull/392
• Fixed broken links by @buvnswrn in https://github.com/NVIDIA/NeMo-Guardrails/pull/670
• Feature/colang 2/standard library updates by @schuellc-nvidia in https://github.com/NVIDIA/NeMo-Guardrails/pull/669
• Fix: hallucination-check by @Pouyanpi in https://github.com/NVIDIA/NeMo-Guardrails/pull/679
• Add the `relevant_chunks` to the gpt-3.5 general prompt template. by @drazvan in https://github.com/NVIDIA/NeMo-Guardrails/pull/678
• Feature/colang 2/flow event bugfix by @schuellc-nvidia in https://github.com/NVIDIA/NeMo-Guardrails/pull/672
• Feat/content safety by @Pouyanpi in https://github.com/NVIDIA/NeMo-Guardrails/pull/674
• feat(embeddings): add search threshold to BasicEmbeddingsIndex by @Pouyanpi in https://github.com/NVIDIA/NeMo-Guardrails/pull/620
• + 49 more

✨ New Contributors

• @saradiazdelser made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/392
• @buvnswrn made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/670
• @vpr1995 made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/695
• @AshishSardana made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/572
• @sklinglernv made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/732
• @radinshayanfar made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/748
• @kauabh made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/727
• Full Changelog: https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.9.1.1...v0.10.0

🐛 Fixed

• [#650](https://github.com/NVIDIA/NeMo-Guardrails/pull/650) Fix gpt-3.5-turbo-instruct prompts #651.
• Full Changelog: https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.9.1...v0.9.1.1

✨ Added

• Colang version [2.0-beta.2](https://github.com/NVIDIA/NeMo-Guardrails/blob/v0.9.1/CHANGELOG-Colang.md#20-beta2---unreleased)
• [#370](https://github.com/NVIDIA/NeMo-Guardrails/pull/370) Add Got It AI's Truthchecking service for RAG applications by @mlmonk.
• [#543](https://github.com/NVIDIA/NeMo-Guardrails/pull/543) Integrating AutoAlign's guardrail library with NeMo Guardrails by @abhijitpal1247.
• [#566](https://github.com/NVIDIA/NeMo-Guardrails/pull/566) Autoalign factcheck examples by @abhijitpal1247.
• [#518](https://github.com/NVIDIA/NeMo-Guardrails/pull/518) Docs: add example config for using models with ollama by @vedantnaik19.
• [#538](https://github.com/NVIDIA/NeMo-Guardrails/pull/538) Support for `--default-config-id` in the server.
• [#539](https://github.com/NVIDIA/NeMo-Guardrails/pull/539) Support for `LLMCallException`.
• [#548](https://github.com/NVIDIA/NeMo-Guardrails/pull/548) Support for custom embedding models.
• + 3 more

📋 Changed

• [#597](https://github.com/NVIDIA/NeMo-Guardrails/pull/597) Make UUID generation predictable in debug-mode.
• [#603](https://github.com/NVIDIA/NeMo-Guardrails/pull/603) Improve chat cli logging.
• [#551](https://github.com/NVIDIA/NeMo-Guardrails/pull/551) Upgrade to Langchain 0.2.x by @nicoloboschi.
• [#611](https://github.com/NVIDIA/NeMo-Guardrails/pull/611) Change default templates.
• [#545](https://github.com/NVIDIA/NeMo-Guardrails/pull/545) NVIDIA API Catalog and NIM documentation update.
• [#463](https://github.com/NVIDIA/NeMo-Guardrails/pull/463) Do not store pip cache during docker build by @don-attilio.
• [#629](https://github.com/NVIDIA/NeMo-Guardrails/pull/629) Move community docs to separate folder.
• [#647](https://github.com/NVIDIA/NeMo-Guardrails/pull/647) Documentation updates.
• + 1 more

🐛 Fixed

• [#482](https://github.com/NVIDIA/NeMo-Guardrails/pull/482) Update README.md by @curefatih.
• [#530](https://github.com/NVIDIA/NeMo-Guardrails/pull/530) Improve the test serialization test to make it more robust.
• [#570](https://github.com/NVIDIA/NeMo-Guardrails/pull/570) Add support for FacialGestureBotAction by @elisam0.
• [#550](https://github.com/NVIDIA/NeMo-Guardrails/pull/550) Fix issue #335 - make import errors visible.
• [#547](https://github.com/NVIDIA/NeMo-Guardrails/pull/547) Fix LLMParams bug and add unit tests (fixes #158).
• [#537](https://github.com/NVIDIA/NeMo-Guardrails/pull/537) Fix directory traversal bug.
• [#536](https://github.com/NVIDIA/NeMo-Guardrails/pull/536) Fix issue #304 NeMo Guardrails packaging.
• [#539](https://github.com/NVIDIA/NeMo-Guardrails/pull/539) Fix bug related to the flow abort logic in Colang 1.0 runtime.
• + 8 more

✨ New Contributors

• @mlmonk made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/370
• @rgstephens made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/504
• @abhijitpal1247 made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/543
• @nicoloboschi made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/551
• @elisam0 made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/570
• @vedantnaik19 made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/518
• @cyun9601 made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/486
• @don-attilio made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/463
• + 2 more

📋 Changes

• A more powerful flow engine supporting multiple parallel flows and advanced pattern matching over the stream of events.
• Adoption of terminology and syntax akin to Python to reduce the learning curve for new developers.
• A standard library and an import mechanism to streamline development.
• Explicit entry point through the `main` flow and explicit activation of flows.
• Smaller set of core abstractions: flows, events, and actions.
• The new _generation operator_ (`...`).
• Asynchronous action execution.

📋 Changed

• [#453](https://github.com/NVIDIA/NeMo-Guardrails/pull/453) Update documentation for NVIDIA API Catalog example.

🐛 Fixed

• [#382](https://github.com/NVIDIA/NeMo-Guardrails/pull/382) Fix issue with `lowest_temperature` in self-check and hallucination rails.
• [#454](https://github.com/NVIDIA/NeMo-Guardrails/pull/454) Redo fix for #385.
• [#442](https://github.com/NVIDIA/NeMo-Guardrails/pull/442) Fix README type by @dileepbapat.

✨ New Contributors

• @dileepbapat made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/442
• Full Changelog: https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.8.2...v0.8.3

✨ Added

• [#402](https://github.com/NVIDIA/NeMo-Guardrails/pull/402) Integrate Vertex AI Models into Guardrails by @aishwaryap.
• [#403](https://github.com/NVIDIA/NeMo-Guardrails/pull/403) Add support for NVIDIA AI Endpoints by @patriciapampanelli
• [#396](https://github.com/NVIDIA/NeMo-Guardrails/pull/396) Docs/examples nv ai foundation models.
• [#438](https://github.com/NVIDIA/NeMo-Guardrails/pull/438) Add research roadmap documentation.

📋 Changed

• [#389](https://github.com/NVIDIA/NeMo-Guardrails/pull/389) Expose the `verbose` parameter through `RunnableRails` by @d-mariano.
• [#415](https://github.com/NVIDIA/NeMo-Guardrails/pull/415) Enable `print(...)` and `log(...)`.
• [#389](https://github.com/NVIDIA/NeMo-Guardrails/pull/389) Expose verbose arg in RunnableRails by @d-mariano.
• [#414](https://github.com/NVIDIA/NeMo-Guardrails/pull/414) Feature/colang march release.
• [#416](https://github.com/NVIDIA/NeMo-Guardrails/pull/416) Refactor and improve the verbose/debug mode.
• [#418](https://github.com/NVIDIA/NeMo-Guardrails/pull/418) Feature/colang flow context sharing.
• [#425](https://github.com/NVIDIA/NeMo-Guardrails/pull/425) Feature/colang meta decorator.
• [#427](https://github.com/NVIDIA/NeMo-Guardrails/pull/427) Feature/colang single flow activation.
• + 4 more

🐛 Fixed

• [#412](https://github.com/NVIDIA/NeMo-Guardrails/pull/412) Fix #411 - explain rails not working for chat models.
• [#413](https://github.com/NVIDIA/NeMo-Guardrails/pull/413) Typo fix: Comment in llm_flows.co by @habanoz.
• [#420](https://github.com/NVIDIA/NeMo-Guardrails/pull/430) Fix typo for hallucination message.

✨ New Contributors

• @aishwaryap made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/402
• @patriciapampanelli made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/403
• @habanoz made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/413
• @d-mariano made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/389
• Full Changelog: https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.8.1...v0.9.0

✨ Added

• [#377](https://github.com/NVIDIA/NeMo-Guardrails/pull/377) Add example for streaming from custom action.

📋 Changed

• [#380](https://github.com/NVIDIA/NeMo-Guardrails/pull/380) Update installation guide for OpenAI usage.
• [#401](https://github.com/NVIDIA/NeMo-Guardrails/pull/401) Replace YAML import with new import statement in multi-modal example.

🐛 Fixed

• [#398](https://github.com/NVIDIA/NeMo-Guardrails/pull/398) Colang parser fixes and improvements.
• [#394](https://github.com/NVIDIA/NeMo-Guardrails/pull/394) Fixes and improvements for Colang 2.0 runtime.
• [#381](https://github.com/NVIDIA/NeMo-Guardrails/pull/381) Fix typo by @serhatgktp.
• [#379](https://github.com/NVIDIA/NeMo-Guardrails/pull/379) Fix missing prompt in verbose mode for chat models.
• [#400](https://github.com/NVIDIA/NeMo-Guardrails/pull/400) Fix Authorization header showing up in logs for NeMo LLM.
• Full Changelog: https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.8.0...v0.8.1

✨ Added

• [#292](https://github.com/NVIDIA/NeMo-Guardrails/pull/292) [Jailbreak heuristics](./docs/user_guides/guardrails-library.md#jailbreak-detection-heuristics) by @erickgalinkin.
• [#256](https://github.com/NVIDIA/NeMo-Guardrails/pull/256) Support [generation options](./docs/user_guides/advanced/generation-options.md).
• [#307](https://github.com/NVIDIA/NeMo-Guardrails/pull/307) Added support for multi-config api calls by @makeshn.
• [#293](https://github.com/NVIDIA/NeMo-Guardrails/pull/293) Adds configurable stop tokens by @zmackie.
• [#334](https://github.com/NVIDIA/NeMo-Guardrails/pull/334) Colang 2.0 - Preview by @schuellc.
• [#208](https://github.com/NVIDIA/NeMo-Guardrails/pull/208) Implement cache embeddings (resolves #200) by @Pouyanpi.
• [#331](https://github.com/NVIDIA/NeMo-Guardrails/pull/331) Huggingface pipeline streaming by @trebedea.
• Documentation:
• + 6 more

📋 Changed

• [#309](https://github.com/NVIDIA/NeMo-Guardrails/pull/309) Change the paper citation from ArXiV to EMNLP 2023 by @manuelciosici
• [#319](https://github.com/NVIDIA/NeMo-Guardrails/pull/319) Enable embeddings model caching.
• [#267](https://github.com/NVIDIA/NeMo-Guardrails/pull/267) Make embeddings computing async and add support for batching.
• [#281](https://github.com/NVIDIA/NeMo-Guardrails/pull/281) Follow symlinks when building knowledge base by @piotrm0.
• [#280](https://github.com/NVIDIA/NeMo-Guardrails/pull/280) Add more information to results of `retrieve_relevant_chunks` by @piotrm0.
• [#332](https://github.com/NVIDIA/NeMo-Guardrails/pull/332) Update docs for batch embedding computations.
• [#244](https://github.com/NVIDIA/NeMo-Guardrails/pull/244) Docs/edit getting started by @DougAtNvidia.
• [#333](https://github.com/NVIDIA/NeMo-Guardrails/pull/333) Follow-up to PR 244.
• + 1 more

🐛 Fixed

• [#286](https://github.com/NVIDIA/NeMo-Guardrails/pull/286) Fixed #285 - using the same evaluation set given a random seed for topical rails by @trebedea.
• [#336](https://github.com/NVIDIA/NeMo-Guardrails/pull/336) Fix #320. Reuse the asyncio loop between sync calls.
• [#337](https://github.com/NVIDIA/NeMo-Guardrails/pull/337) Fix stats gathering in a parallel async setup.
• [#342](https://github.com/NVIDIA/NeMo-Guardrails/pull/342) Fixes OpenAI embeddings support.
• [#346](https://github.com/NVIDIA/NeMo-Guardrails/pull/346) Fix issues with KB embeddings cache, bot intent detection and config ids validator logic.
• [#349](https://github.com/NVIDIA/NeMo-Guardrails/pull/349) Fix multi-config bug, asyncio loop issue and cache folder for embeddings.
• [#350](https://github.com/NVIDIA/NeMo-Guardrails/pull/350) Fix the incorrect logging of an extra dialog rail.
• [#358](https://github.com/NVIDIA/NeMo-Guardrails/pull/358) Fix Openai embeddings async support.
• + 3 more

✨ New Contributors

• @manuelciosici made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/309
• @erickgalinkin made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/292
• @trebedea made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/286
• @piotrm0 made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/281
• @Pouyanpi made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/208
• @niels-garve made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/311
• @zmackie made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/293
• @DougAtNvidia made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/244
• + 3 more

📋 What's Changed

• Replace SentenceTransformers with FastEmbed by @drazvan in https://github.com/NVIDIA/NeMo-Guardrails/pull/288
• Full Changelog: https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.7.0...v0.7.1

✨ Added

• [#254](https://github.com/NVIDIA/NeMo-Guardrails/pull/254) Support for [Llama Guard input and output content moderation](./docs/user_guides/guardrails-library.md#llama-guard-based-content-moderation).
• [#253](https://github.com/NVIDIA/NeMo-Guardrails/pull/253) Support for [server-side threads](./docs/user_guides/server-guide.md#threads).
• [#235](https://github.com/NVIDIA/NeMo-Guardrails/pull/235) Improved [LangChain integration](docs/user_guides/langchain/langchain-integration.md) through `RunnableRails`.
• [#190](https://github.com/NVIDIA/NeMo-Guardrails/pull/190) Add [example](./examples/notebooks/generate_events_and_streaming.ipynb) for using `generate_events_async` with streaming.
• Support for Python 3.11.

📋 Changed

• [#240](https://github.com/NVIDIA/NeMo-Guardrails/pull/240) Switch to pyproject.
• [#276](https://github.com/NVIDIA/NeMo-Guardrails/pull/276) Upgraded Typer to 0.9.

🐛 Fixed

• [#239](https://github.com/NVIDIA/NeMo-Guardrails/pull/239) Fixed logging issue where `verbose=true` flag did not trigger expected log output.
• [#228](https://github.com/NVIDIA/NeMo-Guardrails/pull/228) Fix docstrings for various functions.
• [#242](https://github.com/NVIDIA/NeMo-Guardrails/pull/242) Fix Azure LLM support.
• [#225](https://github.com/NVIDIA/NeMo-Guardrails/pull/225) Fix annoy import, to allow using without.
• [#209](https://github.com/NVIDIA/NeMo-Guardrails/pull/209) Fix user messages missing from prompt.
• [#261](https://github.com/NVIDIA/NeMo-Guardrails/pull/261) Fix small bug in `print_llm_calls_summary`.
• [#252](https://github.com/NVIDIA/NeMo-Guardrails/pull/252) Fixed duplicate loading for the default config.
• Fixed the dependencies pinning, allowing a wider range of dependencies versions.
• + 1 more

✨ New Contributors

• @spehl-max made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/239
• @rajveer43 made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/228
• @smartestrobotdai made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/242
• @prasoonvarshney made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/269
• @eneadodi made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/276
• @baggiponte made their first contribution in https://github.com/NVIDIA/NeMo-Guardrails/pull/240
• Full Changelog: https://github.com/NVIDIA/NeMo-Guardrails/compare/v0.6.1...v0.7.0

✨ Added

• Support for `--version` flag in the CLI.

📋 Changed

• Upgraded `langchain` to `0.0.352`.
• Upgraded `httpx` to `0.24.1`.
• Replaced deprecated `text-davinci-003` model with `gpt-3.5-turbo-instruct`.

🐛 Fixed

• [#191](https://github.com/NVIDIA/NeMo-Guardrails/pull/191): Fix chat generation chunk issue.

📋 Changes

• Better configuration and support for input, output, dialog, retrieval, and execution rails.
• Ability to reduce the overall latency using `single_call` mode or `embeddings_only` mode for dialog rails.
• Support for streaming.
• First version of the Guardrails Library.
• Fast fact-checking using AlignScore.
• Updated Getting Started guide.
• Docker image for easy deployment.

✨ Added

• Support for [explicit definition](./docs/user_guides/configuration-guide.md#guardrails-definitions) of input/output/retrieval rails.
• Support for [custom tasks and their prompts](docs/user_guides/advanced/prompt-customization.md#custom-tasks-and-prompts).
• Support for fact-checking [using AlignScore](./docs/user_guides/guardrails-library.md#alignscore).
• Support for [NeMo LLM Service](./docs/user_guides/configuration-guide.md#nemo-llm-service) as an LLM provider.
• Support for making a single LLM call for both the guardrails process and generating the response (by setting `rails.dialog.single_call.enabled` to `True`).
• Support for [sensitive data detection](./docs/user_guides/guardrails-library.md#sensitive-data-detection) guardrails using Presidio.
• [Example](./examples/configs/llm/hf_pipeline_llama2) using NeMo Guardrails with the LLaMa2-13B model.
• [Dockerfile](./Dockerfile) for building a Docker image.
• + 8 more

📋 Changed

• Allow context data directly in the `/v1/chat/completion` using messages with the type `"role"`.
• Allow calling a subflow whose name is in a variable, e.g. `do $some_name`.
• Allow using actions which are not `async` functions.
• Disabled pretty exceptions in CLI.
• Upgraded dependencies.
• Updated the [Getting Started Guide](./docs/getting_started).
• Main [README](./README.md) now provides more details.
• Merged original examples into a single [ABC Bot](./examples/bots/abc) and removed the original ones.
• + 1 more

🐛 Fixed

• Fix going over the maximum prompt length using the `max_length` attribute in [Prompt Templates](./docs/user_guides/advanced/prompt-customization.md#prompt-templates).
• Fixed problem with `nest_asyncio` initialization.
• [#144](https://github.com/NVIDIA/NeMo-Guardrails/pull/144) Fixed TypeError in logging call.
• [#121](https://github.com/NVIDIA/NeMo-Guardrails/pull/109) Detect chat model using openai engine.
• [#109](https://github.com/NVIDIA/NeMo-Guardrails/pull/109) Fixed minor logging issue.
• Parallel flow support.
• Fix `HuggingFacePipeline` bug related to LangChain version upgrade.

✨ Added

• Support for [custom configuration data](./docs/user_guide/configuration-guide.md#custom-data).
• Example for using [custom LLM and multiple KBs](./examples/multi_kb/README.md)
• Support for [`PROMPTS_DIR`](./docs/user_guide/advanced/prompt-customization.md#prompt-configuration).
• [#101](https://github.com/NVIDIA/NeMo-Guardrails/pull/101) Support for [using OpenAI embeddings](./docs/user_guide/configuration-guide.md#the-embeddings-model) models in addition to SentenceTransformers.
• First set of end-to-end QA tests for the example configurations.
• Support for configurable [embedding search providers](./docs/user_guide/advanced/embedding-search-providers.md)

📋 Changed

• Moved to using `nest_asyncio` for [implementing the blocking API](./docs/user_guide/advanced/nested-async-loop.md). Fixes [#3](https://github.com/NVIDIA/NeMo-Guardrails/issues/3) and [#32](https://github.com/NVIDIA/NeMo-Guardrails/issues/32).
• Improved event property validation in `new_event_dict`.
• Refactored imports to allow installing from source without Annoy/SentenceTransformers (would need a custom embedding search provider to work).

🐛 Fixed

• Fixed when the `init` function from `config.py` is called to allow custom LLM providers to be registered inside.
• [#93](https://github.com/NVIDIA/NeMo-Guardrails/pull/93): Removed redundant `hasattr` check in `nemoguardrails/llm/params.py`.
• [#91](https://github.com/NVIDIA/NeMo-Guardrails/issues/91): Fixed how default context variables are initialized.

✨ Added

• [Event-based API](./docs/user_guide/advanced/event-based-api.md) for guardrails.
• Support for message with type "event" in [`LLMRails.generate_async`](./docs/api/nemoguardrails.rails.llm.llmrails.md#method-llmrailsgenerate_async).
• Support for [bot message instructions](docs/user_guide/advanced/bot-message-instructions.md).
• Support for [using variables inside bot message definitions](./docs/user_guide/colang-language-syntax-guide.md#bot-messages-with-variables).
• Support for `vicuna-7b-v1.3` and `mpt-7b-instruct`.
• Topical evaluation results for `vicuna-7b-v1.3` and `mpt-7b-instruct`.
• Support to use different models for different LLM tasks.
• Support for [red-teaming](docs/user_guide/advanced/red-teaming.md) using challenges.
• + 3 more

📋 Changed

• Changed the naming of the internal events to align to the upcoming UMIM spec (Unified Multimodal Interaction Management).
• If there are no user message examples, the bot messages examples lookup is disabled as well.

🐛 Fixed

• [#58](https://github.com/NVIDIA/NeMo-Guardrails/issues/58): Fix install on Mac OS 13.
• [#55](https://github.com/NVIDIA/NeMo-Guardrails/issues/55): Fix bug in example causing config.py to crash on computers with no CUDA-enabled GPUs.
• Fixed the model name initialization for LLMs that use the `model` kwarg.
• Fixed the Cohere prompt templates.
• [#55](https://github.com/NVIDIA/NeMo-Guardrails/issues/83): Fix bug related to LangChain callbacks initialization.
• Fixed generation of "..." on value generation.
• Fixed the parameters type conversion when invoking actions from colang (previously everything was string).
• Fixed `model_kwargs` property for the `WrapperLLM`.
• + 2 more

✨ Added

• Support for defining [subflows](docs/user_guide/colang-language-syntax-guide.md#subflows).
• Improved support for [customizing LLM prompts](docs/user_guide/advanced/prompt-customization.md)
• Support for using filters to change how variables are included in a prompt template.
• Output parsers for prompt templates.
• The `verbose_v1` formatter and output parser to be used for smaller models that don't understand Colang very well in a few-shot manner.
• Support for including context variables in prompt templates.
• Support for chat models i.e. prompting with a sequence of messages.
• Experimental support for allowing the LLM to generate [multi-step flows](docs/user_guide/configuration-guide.md#multi-step-generation).
• + 7 more

📋 Changed

• The core templates now use Jinja2 as the rendering engines.
• Improved the internal prompting architecture, now using an LLM Task Manager.

🐛 Fixed

• Fixed bug related to invoking a chain with multiple output keys.
• Fixed bug related to tracking the output stats.
• #51: Bug fix - avoid str concat with None when logging user_intent.
• #54: Fix UTF-8 encoding issue and add embedding model configuration.