Skip to content
This repository was archived by the owner on May 20, 2026. It is now read-only.

change vsc promptD#5117

Open
thirteenflt wants to merge 1 commit into
microsoft:mainfrom
thirteenflt:yt_dev
Open

change vsc promptD#5117
thirteenflt wants to merge 1 commit into
microsoft:mainfrom
thirteenflt:yt_dev

Conversation

@thirteenflt
Copy link
Copy Markdown
Contributor

No description provided.

Copilot AI review requested due to automatic review settings May 15, 2026 22:30
@vs-code-engineering
Copy link
Copy Markdown
Contributor

📬 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
Copy Markdown
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 VSC Model D’s agent prompt instructions around commentary-channel intermediary updates and channel-order behavior.

Changes:

  • Adds a more detailed intermediary_updates section with progress-report and rationale-update guidance.
  • Updates commentary channel instructions to reference the new intermediary update rules.
  • Replaces the previous shorter intermediary update guidance with a first-commentary-specific rule.
Show a summary per file
File Description
src/extension/prompts/node/agent/vscModelPrompts.tsx Revises VSC Model D prompt text for commentary updates and channel usage.

Copilot's findings

  • Files reviewed: 1/1 changed files
  • Comments generated: 5

Comment on lines +627 to +632
Kind 2 — Rationale updates (longer):<br />
- Once you have gathered enough context to reason about the problem, you are STRONGLY encouraged to share your thinking and rationale with the user — for example, before forming a plan, choosing an approach, diagnosing a bug, or making a non-trivial decision.<br />
- This MUST be written in fluent, natural language as you would explain your thought process to a colleague. It must NOT resemble internal chain-of-thought, bullet-style scratch notes, or model reasoning traces.<br />
- Each rationale update MUST NOT exceed 200 tokens.<br />
- When you do not yet have enough context, or rationale is not needed at a particular step, use a Kind 1 progress report instead.<br />
- After you have sufficient context and the work is substantial, you may provide a longer plan — this is the only update that may exceed 2 sentences and can contain formatting.<br />
- Combining multiple line references in one link: [file.ts#L10-L12, L20](file.ts#L10-L12, L20)<br />
</Tag>
<Tag name='intermediary_updates'>
All intermediary updates go to the `commentary` channel. They are user-facing and must NOT be final answers. There are two kinds of intermediary update; choose the appropriate one at each step.<br />
- Before performing file edits of any kind, you provide updates explaining what edits you are making.<br />
- As you are thinking, you very frequently provide updates even if not taking any actions, informing the user of your progress. Do not accumulate long uninterrupted internal thinking without a commentary update. If your thinking exceeds 256 cumulative words since the last user-facing update, send a commentary update before continuing. If thinking continues, send additional commentary updates at least every further 256 words.<br />
- Tone of your updates MUST match your personality.<br />
- The first commentary should not begin updates with conversational interjections or meta commentary. Avoid openers such as acknowledgements ("Done —", "Got it", "Great question,") or framing phrases. Do not use starters like "Got it -" or "Understood -".<br />
- After you have sufficient context and the work is substantial, you may provide a longer plan — this is the only update that may exceed 2 sentences and can contain formatting.<br />
<br />
General rules for both kinds:<br />
- Every commentary message must be written in complete, fluent, natural-language sentences that form grammatically complete thoughts. Bare noun phrases, keyword fragments, isolated labels, or telegraphic shorthand are not acceptable (e.g., "File structure." or "Auth module, config, tests." are disallowed).<br />
All intermediary updates go to the `commentary` channel. They are user-facing and must NOT be final answers. There are two kinds of intermediary update; choose the appropriate one at each step.<br />
<br />
Kind 1 — Progress reports (short):<br />
- 1-3 sentence updates that tell the user what you just did, found, or will do next.<br />
@alexdima
Copy link
Copy Markdown
Member

Thanks for the contribution! This repository has been archived because the project has moved into the main VS Code repository.

Could you please reopen/recreate this PR against:
https://github.com/microsoft/vscode/tree/main/extensions/copilot

We’ll continue reviewing contributions there. Thanks!

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

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants