mirror of
https://ghfast.top/https://github.com/discourse/discourse-ai.git
synced 2026-07-16 11:36:23 +08:00
This is a significant PR that introduces AI Artifacts functionality to the discourse-ai plugin along with several other improvements. Here are the key changes: 1. AI Artifacts System: - Adds a new `AiArtifact` model and database migration - Allows creation of web artifacts with HTML, CSS, and JavaScript content - Introduces security settings (`strict`, `lax`, `disabled`) for controlling artifact execution - Implements artifact rendering in iframes with sandbox protection - New `CreateArtifact` tool for AI to generate interactive content 2. Tool System Improvements: - Adds support for partial tool calls, allowing incremental updates during generation - Better handling of tool call states and progress tracking - Improved XML tool processing with CDATA support - Fixes for tool parameter handling and duplicate invocations 3. LLM Provider Updates: - Updates for Anthropic Claude models with correct token limits - Adds support for native/XML tool modes in Gemini integration - Adds new model configurations including Llama 3.1 models - Improvements to streaming response handling 4. UI Enhancements: - New artifact viewer component with expand/collapse functionality - Security controls for artifact execution (click-to-run in strict mode) - Improved dialog and response handling - Better error management for tool execution 5. Security Improvements: - Sandbox controls for artifact execution - Public/private artifact sharing controls - Security settings to control artifact behavior - CSP and frame-options handling for artifacts 6. Technical Improvements: - Better post streaming implementation - Improved error handling in completions - Better memory management for partial tool calls - Enhanced testing coverage 7. Configuration: - New site settings for artifact security - Extended LLM model configurations - Additional tool configuration options This PR significantly enhances the plugin's capabilities for generating and displaying interactive content while maintaining security and providing flexible configuration options for administrators.
130 lines
4 KiB
Ruby
130 lines
4 KiB
Ruby
# frozen_string_literal: true
|
|
|
|
module DiscourseAi
|
|
module Completions
|
|
module Dialects
|
|
class XmlTools
|
|
def initialize(tools)
|
|
@raw_tools = tools
|
|
end
|
|
|
|
def translated_tools
|
|
raw_tools.reduce(+"") do |tools, function|
|
|
parameters = +""
|
|
if function[:parameters].present?
|
|
function[:parameters].each do |parameter|
|
|
parameters << <<~PARAMETER
|
|
<parameter>
|
|
<name>#{parameter[:name]}</name>
|
|
<type>#{parameter[:type]}</type>
|
|
<description>#{parameter[:description]}</description>
|
|
<required>#{parameter[:required]}</required>
|
|
PARAMETER
|
|
if parameter[:enum]
|
|
parameters << "<options>#{parameter[:enum].join(",")}</options>\n"
|
|
end
|
|
parameters << "</parameter>\n"
|
|
end
|
|
end
|
|
|
|
tools << <<~TOOLS
|
|
<tool_description>
|
|
<tool_name>#{function[:name]}</tool_name>
|
|
<description>#{function[:description]}</description>
|
|
<parameters>
|
|
#{parameters}</parameters>
|
|
</tool_description>
|
|
TOOLS
|
|
end
|
|
end
|
|
|
|
def instructions
|
|
return "" if raw_tools.blank?
|
|
|
|
@instructions ||=
|
|
begin
|
|
has_arrays =
|
|
raw_tools.any? { |tool| tool[:parameters]&.any? { |p| p[:type] == "array" } }
|
|
|
|
(<<~TEXT).strip
|
|
#{tool_preamble(include_array_tip: has_arrays)}
|
|
<tools>
|
|
#{translated_tools}</tools>
|
|
TEXT
|
|
end
|
|
end
|
|
|
|
def from_raw_tool(raw_message)
|
|
(<<~TEXT).strip
|
|
<function_results>
|
|
<result>
|
|
<tool_name>#{raw_message[:name] || raw_message[:id]}</tool_name>
|
|
<json>
|
|
#{raw_message[:content]}
|
|
</json>
|
|
</result>
|
|
</function_results>
|
|
TEXT
|
|
end
|
|
|
|
def from_raw_tool_call(raw_message)
|
|
parsed = JSON.parse(raw_message[:content], symbolize_names: true)
|
|
parameters = +""
|
|
|
|
if parsed[:arguments]
|
|
parameters << "<parameters>\n"
|
|
parsed[:arguments].each { |k, v| parameters << "<#{k}>#{v}</#{k}>\n" }
|
|
parameters << "</parameters>\n"
|
|
end
|
|
|
|
(<<~TEXT).strip
|
|
<function_calls>
|
|
<invoke>
|
|
<tool_name>#{raw_message[:name] || parsed[:name]}</tool_name>
|
|
#{parameters}</invoke>
|
|
</function_calls>
|
|
TEXT
|
|
end
|
|
|
|
private
|
|
|
|
attr_reader :raw_tools
|
|
|
|
def tool_preamble(include_array_tip: true)
|
|
array_tip =
|
|
if include_array_tip
|
|
<<~TEXT
|
|
If a parameter type is an array, return an array of values. For example:
|
|
<$PARAMETER_NAME>["one","two","three"]</$PARAMETER_NAME>
|
|
TEXT
|
|
else
|
|
""
|
|
end
|
|
|
|
<<~TEXT
|
|
In this environment you have access to a set of tools you can use to answer the user's question.
|
|
You may call them like this.
|
|
|
|
<function_calls>
|
|
<invoke>
|
|
<tool_name>$TOOL_NAME</tool_name>
|
|
<parameters>
|
|
<$PARAMETER_NAME>$PARAMETER_VALUE</$PARAMETER_NAME>
|
|
...
|
|
</parameters>
|
|
</invoke>
|
|
</function_calls>
|
|
#{array_tip}
|
|
If you wish to call multiple function in one reply, wrap multiple <invoke>
|
|
block in a single <function_calls> block.
|
|
|
|
- Always prefer to lead with tool calls, if you need to execute any.
|
|
- Avoid all niceties prior to tool calls, Eg: "Let me look this up for you.." etc.
|
|
- DO NOT encode HTML entities in tool calls. You may use <![CDATA[...]]> for encoding if required.
|
|
Here are the complete list of tools available:
|
|
TEXT
|
|
end
|
|
end
|
|
end
|
|
end
|
|
end
|