Skip to content

Add Frozen prompt for 0129 to Prompt A for VSC Chat model#3452

Open
ashatabak786 wants to merge 8 commits intomicrosoft:mainfrom
AbdelrahmanAbouelenin:atabakashfaq/add0129_promptA
Open

Add Frozen prompt for 0129 to Prompt A for VSC Chat model#3452
ashatabak786 wants to merge 8 commits intomicrosoft:mainfrom
AbdelrahmanAbouelenin:atabakashfaq/add0129_promptA

Conversation

@ashatabak786
Copy link
Contributor

Change PromptA to 0129 version
Add swe_vsc_derisk_ttfc0-s90 hash to PromptA

@ashatabak786 ashatabak786 marked this pull request as ready for review February 4, 2026 22:42
@vs-code-engineering
Copy link

vs-code-engineering bot commented Feb 4, 2026

📬 CODENOTIFY

The following users are being notified based on files changed in this PR:

@karthiknadig

Matched files:

  • src/extension/prompts/node/agent/vscModelPrompts.tsx

@eleanorjboyd

Matched files:

  • src/extension/prompts/node/agent/vscModelPrompts.tsx

Copy link
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

This PR updates the VSC Model A prompt to a new "0129" version by replacing the entire system prompt content while maintaining the same structure. The changes represent a significant shift in prompting strategy, removing planning/todo list instructions and response translation rules, while adding new guidance around parallel tool usage, response formatting requirements, and assistant message ordering.

Changes:

  • Replaced planning/todo list instructions with parallel tool use guidance emphasizing multi_tool_use for efficiency
  • Added explicit final_first_requirement and commentary_first_requirement tags to control assistant message ordering
  • Created new VSCModelReminderInstructionsA class with additional constraints referencing the new requirements
  • Removed imports for ResponseTranslationRules, McpToolInstructions, and NotebookInstructions that are no longer used
Comments suppressed due to low confidence (3)

src/extension/prompts/node/agent/vscModelPrompts.tsx:110

  • The response text "sure I am here to help." should be "Sure, I am here to help." with a capital "S" at the beginning and a comma after "Sure" for proper grammar. Additionally, there's a double period ".." which should be a single period.
				if the request is ambiguous, respond with "sure I am here to help.".<br />

src/extension/prompts/node/agent/vscModelPrompts.tsx:116

  • There are two consecutive blank lines (115-116) before the closing tag. This appears to be unnecessary whitespace. Consider removing one blank line for consistency with the rest of the codebase.
    src/extension/prompts/node/agent/vscModelPrompts.tsx:24
  • There's a grammatical error with a duplicated word "If". The sentence reads "You are encouraged to call functions in parallel if If you think..." - the second "If" should be removed or the sentence should be restructured.
				You are encouraged to call functions in parallel if If you think running multiple tools can answer the user's question to maximize efficiency by parallelizing independent operations. This reduces latency and provides faster responses to users.<br />

Comment on lines +104 to +110
If the answer is direct and needs no tools or multi-step work (e.g. User say hello), respond with ONE final message only. No commentary or analysis messages are needed. That is, you should only send one message, the final answer.<br />
You CANNOT call commentary and then final right after that.<br />
</Tag>
<Tag name='commentary_first_requirement'>
If not satisfying the final_first_requirement, you should ALWAYS obey this requirement: before starting any analysis or tool call, send an initial commentary-channel message that is at most two sentences (prefer one).<br />
It must restate the user's clear request while acknowledging you will handle it.<br />
if the request is ambiguous, respond with "sure I am here to help.".<br />
Copy link

Copilot AI Feb 4, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Grammatical error: "User say hello" should be "User says hello" to maintain subject-verb agreement.

This issue also appears in the following locations of the same file:

  • line 110
  • line 24
Suggested change
If the answer is direct and needs no tools or multi-step work (e.g. User say hello), respond with ONE final message only. No commentary or analysis messages are needed. That is, you should only send one message, the final answer.<br />
You CANNOT call commentary and then final right after that.<br />
</Tag>
<Tag name='commentary_first_requirement'>
If not satisfying the final_first_requirement, you should ALWAYS obey this requirement: before starting any analysis or tool call, send an initial commentary-channel message that is at most two sentences (prefer one).<br />
It must restate the user's clear request while acknowledging you will handle it.<br />
if the request is ambiguous, respond with "sure I am here to help.".<br />
If the answer is direct and needs no tools or multi-step work (e.g. user says hello), respond with ONE final message only. No commentary or analysis messages are needed. That is, you should only send one message, the final answer.<br />
You CANNOT call commentary and then final right after that.<br />
</Tag>
<Tag name='commentary_first_requirement'>
If not satisfying the final_first_requirement, you should ALWAYS obey this requirement: before starting any analysis or tool call, send an initial commentary-channel message that is at most two sentences (prefer one).<br />
It must restate the user's clear request while acknowledging you will handle it.<br />
If the request is ambiguous, respond with "Sure, I am here to help.".<br />

Copilot uses AI. Check for mistakes.
- Read file content → analyze → then edit based on content<br />
- Semantic search for context → wait → then read specific files<br />
<br />
❌ BAD<br />
Copy link

Copilot AI Feb 4, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Inconsistent formatting with VSCModelPromptB. In line 59, the heading is just "❌ BAD" without additional context, while VSCModelPromptB at line 165 has "❌ BAD - Exceeding parallel limits:" which is more descriptive. Consider adding a descriptive label after "BAD" for consistency and clarity.

Suggested change
BAD<br />
BAD - Exceeding parallel limits:<br />

Copilot uses AI. Check for mistakes.
Copy link
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants