Merge branch 'main' into 15-add-data-sources---csv-parquet

CHANGELOG.md

@@ -1,3 +1,9 @@
+## v0.5.0 (2024-09-10)
+
+### Feat
+
+- added env var KONTEXT_COPILOT_DEFAULT_SYS_PROMPT to set differe… (#50)
+
 ## v0.4.0 (2024-09-09)
 
 ### Feat

CODEOWNERS

@@ -0,0 +1 @@
+@kontext-tech/kontext-copilot-reviewer

README.md

@@ -2,6 +2,8 @@
 
 Kontext Copilot is an AI empowered assistant for data analytics that runs on your local computer.
 
+![kontext-copilot-example](https://cdn.kontext.tech/Images/ai/kontext-copilot-light.gif)
+
 ## Prerequisites
 
 -  Python 3.9+
@@ -27,7 +29,9 @@ Kontext Copilot is still at early stage. Please follow the guide below to evalua
 
 [**Get started with Kontext Copilot**](https://kontext.tech/article/1386/get-started-with-kontext-copilot)
 
-![kontext-copilot-example](https://cdn.kontext.tech/Images/ai/kontext-copilot-light.gif)
+## Roadmap
+
+![kontext-copilot-roadmap](https://kontext.tech/api/flex/diagram/diagram-1388)
 
 ## Change logs
 

kontext_copilot/copilot/_prompt_factory.py

@@ -51,7 +51,7 @@ def create_system_prompt(
         logger.info("Creating system prompt for data source: %s", data_source.name)
 
         prompt_model = PromptFactory.get_prompt_template_by_id(
-            "system-prompt-da-metadata"
+            os.getenv("KONTEXT_COPILOT_DEFAULT_SYS_PROMPT", "system-prompt-da-metadata")
         )
 
         logger.debug("Prompt template: %s", prompt_model)

kontext_copilot/copilot/prompts.json

@@ -16,6 +16,14 @@
          "user_input": null,
          "system_defined": true
       },
+      {
+         "id": "system-prompt-da-metadata-refined",
+         "name": "System Prompt - Data Analytics (using Metadata)",
+         "prompt": null,
+         "system_prompt": "./prompts/system-prompt-da-metadata-refined.md",
+         "user_input": null,
+         "system_defined": true
+      },
       {
          "id": "sentiment-analysis",
          "name": "Sentiment Analysis",

kontext_copilot/copilot/prompts/system-prompt-da-metadata-refined.md

@@ -0,0 +1,8 @@
+You are an expert AI Assistant specializing in data analysis. Your task is to generate precise and correct SQL queries or code snippets using the {{$database_type}} dialect.
+
+Here are the database details:
+<database_type>{{$database_type}}</database_type>
+<database_name>{{$database_name}}</database_name>
+<table_metadata>{{$tables_metadata}}</table_metadata>
+
+Only use the provided table metadata to construct correct and executable SQL queries. Ensure your responses strictly adhere to the {{$database_type}} database functions and syntax. If you require additional information or clarification, ask the user specific questions. Always provide SQL queries that can be executed without errors.

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "kontext-copilot"
-version = "0.4.0"
+version = "0.5.0"
 description = "Kontext Copilot is an AI empowered assistant for data analytics that can run on your local computer."
 authors = ["Kontext <[email protected]>"]
 readme = "README.md"