v0.2.18

-added reattach when initializing spool

v0.2.17

-added ollama and openai function output

v0.2.16

-added a /flush macro and a /rehash macro.
-flush flushes context. defaults to all but can set n_last. will eventually add n_start and n_last.
-rehash sends the most recent specified message.

v0.2.15

-added in pipeline examples to readme and built out pipelinerunner functionalities
-fixed saving location for image generation to be in ~/.npcsh/images

v0.2.14

-fixed screencap_tool issues

v0.2.13

-added more python examples to the readme. fixed up some code in order for them to work

v0.2.12

-added substantial updates to the readme, including some more images and example outputs for npcsh prompts and macros
-fixed some issues with image generation for stable diffusion
-fixed some issues with tools not working properly (search, image generation, screen cap seem to work well now)

v0.2.11

-added in file inclusion to spool mode so you can have threaded conversations with raggable text from a pdf.
-you can include an alternative rag_similarity threshold (default is 0.3) by specificying it in the /spool command.
=you can also specify the device (gpu or cpu) tho this isnt yet implemented yet in the rag search.

v0.2.10

-fixed a message duplication in execute_llm_question.
-fixed a model overriding issuee wherein nones were being passed for the provider and model to check llm command because the model override was finding nothing. so now if there is a model override it takes precedent over the env variable defined ones.

v0.2.9

-fixed npc query error