You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: src/i18n/locales/en/common.json
+3-3Lines changed: 3 additions & 3 deletions
Original file line number
Diff line number
Diff line change
@@ -47,7 +47,7 @@
47
47
"create_mcp_json": "Failed to create or open .roo/mcp.json: {{error}}",
48
48
"hmr_not_running": "Local development server is not running, HMR will not work. Please run 'npm run dev' before launching the extension to enable HMR.",
49
49
"retrieve_current_mode": "Error: failed to retrieve current mode from state.",
50
-
"summarization_failed": "Failed to summarize conversation context.",
50
+
"synthesization_failed": "Failed to synthesize conversation context.",
51
51
"failed_delete_repo": "Failed to delete associated shadow repository or branch: {{error}}",
52
52
"failed_remove_directory": "Failed to remove task directory: {{error}}",
53
53
"custom_storage_path_unusable": "Custom storage path \"{{path}}\" is unusable, will use default path",
// Construct the prompt for the summarization model (User Final Refinement)
28
+
// Construct the prompt for the synthesizing model (User Final Refinement)
29
29
constsystemPrompt=`You are a specialized context compression system for Roo-Code, a VS Code extension that enables AI coding agents. Your sole purpose is to condense conversation history while preserving maximum technical context with minimum tokens.
30
30
31
31
**Context Schema:**
32
-
- You are summarizing the MIDDLE portion of a conversation
33
-
- The original system prompt and initial interactions remain intact before your summary
34
-
- Recent conversation turns remain intact after your summary
35
-
- Your summary will be the critical bridge connecting these preserved segments
32
+
- You are synthesizing the MIDDLE portion of a conversation
33
+
- The original system prompt and initial interactions remain intact before your synthesis
34
+
- Recent conversation turns remain intact after your synthesis
35
+
- Your synthesis will be the critical bridge connecting these preserved segments
36
36
37
37
**Content Priorities (Highest to Lowest):**
38
38
1. **Code Context:**
@@ -64,19 +64,19 @@ export class ContextSummarizer {
64
64
- Testing approaches
65
65
66
66
**Output Requirements:**
67
-
- Produce ONLY the summary text with no meta-commentary
67
+
- Produce ONLY the synthesis text with no meta-commentary
68
68
- Use precise, technical language optimized for information density
69
69
- Structure with minimal formatting (use ## for major sections if necessary)
70
70
- Omit pleasantries, acknowledgments, and conversational elements
71
71
- Format sequences of related facts as compact, semicolon-separated phrases
72
72
- Use minimal tokens while maximizing preserved information
73
73
- Prioritize factual over instructional content
74
74
75
-
This summary must enable seamless conversation continuity with no perceived context loss between the earlier and later preserved segments.`
75
+
This synthesis must enable seamless conversation continuity with no perceived context loss between the earlier and later preserved segments.`
76
76
77
77
// Format the messages for the prompt. Simple stringification might be too verbose or lose structure.
78
78
// Let's try a more readable format.
79
-
constformattedMessages=messagesToSummarize
79
+
constformattedMessages=messagesToSynthesize
80
80
.map((msg)=>{
81
81
letcontentText=""
82
82
if(Array.isArray(msg.content)){
@@ -95,52 +95,52 @@ This summary must enable seamless conversation continuity with no perceived cont
95
95
})
96
96
.join("\n\n---\n\n")
97
97
98
-
constuserPrompt=`Please summarize the following conversation turns:\n\n${formattedMessages}`
98
+
constuserPrompt=`Please synthesize the following conversation turns:\n\n${formattedMessages}`
99
99
100
100
try{
101
-
// Use the configured API handler to make the summarization call
102
-
// Note: This uses the main configured model. Consider allowing a specific summarization model.
103
-
// Disable prompt caching for summarization calls? - Currently not directly supported per-call.
101
+
// Use the configured API handler to make the synthesizing call
102
+
// Note: This uses the main configured model. Consider allowing a specific synthesizing model.
103
+
// Disable prompt caching for synthesizing calls? - Currently not directly supported per-call.
104
104
// It will use the handler's configured caching setting.
105
105
conststream=this.apiHandler.createMessage(
106
106
systemPrompt,
107
107
[{role: "user",content: userPrompt}],
108
-
undefined,// No specific cache key for summarization
108
+
undefined,// No specific cache key for synthesizing
0 commit comments