@zimtsui/typechat 0.0.100 → 0.0.101

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -2,13 +2,13 @@
2
2
 
3
3
  ## Motivation
4
4
 
5
- When a LLM outputs structured data in JSON format (e.g., legacy LLM function calling), if there are too many special characters in a string property (e.g., a large LaTeX document, or a complex shell command), the LLM is prone to make mistakes in JSON escaping.
5
+ When an LLM outputs structured data in JSON format (e.g., legacy LLM function calling), if there are too many special characters in a string property (e.g., a large LaTeX document, or a complex shell command), the LLM is prone to make mistakes in JSON escaping.
6
6
 
7
7
  XML Verbatim Channel is designed to avoid escaping in LLM messages.
8
8
 
9
9
  ## Declaration of Channels
10
10
 
11
- The LLM system/user message is expected to declare all available XML Verbatim Channels in the form of
11
+ LLM developer/user messages are expected to declare all available XML Verbatim Channels in the form of
12
12
 
13
13
  <verbatim:declaration name="NAME OF CHANNEL">
14
14
  <verbatim:description>DESCRIPTION OF CHANNEL</verbatim:description>
@@ -26,7 +26,7 @@ The LLM system/user message is expected to declare all available XML Verbatim Ch
26
26
 
27
27
  ## Request through Channels
28
28
 
29
- You can make a request through a channel in the form of
29
+ In an LLM AI message you output, you can make a request through a specified channel in the form of
30
30
 
31
31
  <verbatim:request name="NAME OF CHANNEL">
32
32
  <verbatim:parameter name="NAME OF PARAMETER 1"><![CDATA[VERBATIM ARGUMENT OF THIS PARAMETER]]></verbatim:parameter>
@@ -39,7 +39,7 @@ You can make a request through a channel in the form of
39
39
 
40
40
  ## Response from Channels
41
41
 
42
- The LLM user message may contain the responses to your request in the form of
42
+ The LLM user message following your requests may contain the responses to your requests in the form of
43
43
 
44
44
  <verbatim:response name="NAME OF CHANNEL"><![CDATA[VERBATIM RESPONSE]]></verbatim:response>
45
45
 
@@ -49,9 +49,9 @@ Not all requests have a response.
49
49
 
50
50
  ## Verbatim Quotation
51
51
 
52
- The LLM system/user message may contain verbatim quotations in the form of
52
+ LLM developer/user messages may contain verbatim quotations in the form of
53
53
 
54
- <verbatim:quotation author="AUTHOR OF THIS QUOTATION"><![CDATA[VERBATIM QUOTATION]]></verbatim:quotation>
54
+ <verbatim:quotation author="AUTHOR OF QUOTATION"><![CDATA[VERBATIM QUOTATION]]></verbatim:quotation>
55
55
 
56
56
  The CDATA section may directly contain `]]>`, which is not allowed in standard XML.
57
57
 
@@ -59,6 +59,6 @@ The attribute `author` is optional.
59
59
 
60
60
  ## System Information
61
61
 
62
- The LLM system/user message may contain system information from the AI agent framework in the form of
62
+ LLM developer/user messages may contain system information from the AI agent framework in the form of
63
63
 
64
64
  <verbatim:system></verbatim:system>
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@zimtsui/typechat",
3
- "version": "0.0.100",
3
+ "version": "0.0.101",
4
4
  "exports": {
5
5
  ".": "./build/exports.js",
6
6
  "./assets": "./build/assets.js",