v0.3.0

Features:

• langchain 0.4 (#132) by @barnabasJ (#132)

Bug Fixes:

• turn content into string for chat title by @barnabasJ (#132)

Improvements:

• update ash_oban to 0.5 by @zachdaniel

v0.2.14

Improvements:

• allow multiple full text definitions in vectorize section (#125) by Abdessabour Moutik (#125)

v0.2.9

Improvements:

• mark all fields as required by @zachdaniel

• handle number constraints by @zachdaniel

• Add on_tool_start and on_tool_end callbacks (#96) by @bradleygolden

v0.2.8

Improvements:

• add typed struct example to usage rules & docs by @zachdaniel

v0.2.7

Bug Fixes:

• separate custom_context from llm initialization in ash_ai.gen.chat (#88) by @germanbottosur

v0.2.6

Bug Fixes:

• handle missing user module more gracefully by @zachdaniel

• properly install usage rules by @zachdaniel

v0.2.5

Improvements:

• support sub rules in usage rules tools by @zachdaniel

v0.2.4

Bug Fixes:

• allow for a custom json_processor (#80) by @TwistingTwists

• changed chat-live message history order before adding it to langchain (#78) by srmico

• crash with embedded resource (#77) by @nallwhy

Improvements:

• add documentation for tool private attribute behavior (#81) by marot

• add documentation for tool private attribute behavior by marot

• install usage rules better by @zachdaniel

v0.2.3

Bug Fixes:

• unsafe usage in mdex (#73) by @TwistingTwists

Improvements:

• update usage rules w/ more prompt actions by @zachdaniel

• multi-provider support prerequisite - eliminate open api spex reliance (#64) by KasparKipp

• Support various additional prompt formats (#72) by @TwistingTwists