1.9.1

• Improved ability for dir-assistant to be run concurrently in the same system environment by adding WAL and timeout to SQLite instances.

1.9.0

• Switched from euclidean distance to cosine similarity for artifact relevancy filtering. When upgrading, you will need to run dir-assistant clear.
• Added ARTIFACT_COSINE_CUTOFF and ARTIFACT_COSINE_CGRAG_CUTOFF to exclude artifacts with low cosine similarity.
• Updated support for the latest version of llama-cpp-python.

1.8.1

• COMMIT_TO_GIT now allows absolute filepaths if they are within the CWD

1.8.0

• The embedding index is now separated by model in the database. This means you can switch between embedding models without needing to re-index files.
• Indexing is now parallelized and faster.
• Fixed a bug where the context optimizer was not utilizing the whole context

1.7.0

• Context Pretext Caching
• Separate CGRAG model options

1.6.0

Bug fixes and minor improvements

Full Changelog: 1.5.0...1.6.0

1.5.0

• Added support for models that include a <thinking></thinking> block in their response.
• Added an example script for analyzing stock sentiment on reddit.

1.4.2

Improved Windows support

1.4.0

• Added llama-cpp-python as an optional instead of required dependency downloadable
  with pip install dir-assistant[recommended]
• Official Windows support
• Custom API server connections using the new LiteLLM completion settings config section. This enables
  you to use your own GPU rig with dir-assistant.

1.3.2

• Reverted untested default embed chunk size