Skip to content

Handling special claude model response in github copilot provider #6369

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
lifeizhou-ap merged 2 commits into from
Jan 7, 2026
Merged

Handling special claude model response in github copilot provider #6369

lifeizhou-ap merged 2 commits into from
Jan 7, 2026
+89 −4

Conversation

@lifeizhou-ap
Copy link
Collaborator

@lifeizhou-ap lifeizhou-ap commented Jan 7, 2026 •
edited
Loading

Summary

Why

With the PR #6331, the tool call are no longer triggered when we using github copilot provider with Claude models.

That change in the PR let Goose consume GitHub Copilot’s SSE stream directly. Copilot’s Claude Sonnet stream doesn’t include function.arguments, so tool calls never fire.

Why it was working before PR #6331
Before #6331 we buffered the entire stream inside the provider, so Claude effectively behaved like a non-streaming model and tools worked.

What

  • Remove Claude models from GITHUB_COPILOT_STREAM_MODELS so they go through the completion path again.
  • In GithubCopilotProvider::complete_with_model, call promote_tool_choice() to reorder the completion response by promoting whichever choice contains tool_calls to index 0 before passing it to the shared OpenAI formatter. This special handling is for the Claude models (more explain in the promote_tool_choice function comments.

Type of Change

  • Feature
  • Bug fix
  • Refactor / Code quality
  • Performance improvement
  • Documentation
  • Tests
  • Security fix
  • Build / Release
  • Other (specify below)

AI Assistance

  • This PR was created or reviewed with AI assistance

Testing

Unit test and Manual testing

Copilot AI review requested due to automatic review settings January 7, 2026 09:59
Copilot AI reviewed Jan 7, 2026
View reviewed changes
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 fixes a regression where tool calls stopped working for Claude models when using the GitHub Copilot provider. The issue was introduced by PR #6331 which enabled direct SSE stream consumption, but Copilot's Claude stream doesn't include function.arguments in tool calls.

Key Changes

  • Removes Claude models (claude-sonnet-4, claude-sonnet-4.5, claude-haiku-4.5) from GITHUB_COPILOT_STREAM_MODELS so they use the completion path instead of streaming
  • Adds promote_tool_choice() function to handle GitHub Copilot's behavior of returning tool_calls in non-zero index choices for Claude models
  • Includes unit tests covering both promotion and no-op scenarios

jamadeo
jamadeo approved these changes Jan 7, 2026
View reviewed changes
crates/goose/src/providers/githubcopilot.rs
}
return new_response;
}
}
Copy link
Collaborator

@jamadeo jamadeo Jan 7, 2026

Choose a reason for hiding this comment

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

makes sense to me, but wouldn't it be simpler to do a Vec::sort_by?

also I wonder if this would make sense to have in formats::openai::response_to_message, since that is where we select the first choice

Copy link
Collaborator Author

@lifeizhou-ap lifeizhou-ap Jan 7, 2026 •
edited
Loading

Choose a reason for hiding this comment

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

@jamadeo Thanks for the review!

"sort_by" could reorder all the choices. the current approach only moves the first choice with "tool_calls" to index 0 while preserving the rest.

also I wonder if this would make sense to have in formats::openai::response_to_message, since that is where we select the first choice

Initially I implemented this logic in formats::openai::response_to_message, but later on I moved to this logic to this file to avoid complexity in formats::openai as this is the special case with github_copilot provider. When the other provider has the similar use case, we can move to formats::openai. WDYT? I can create a followup PR if it makes more sense to include this logic in formats::openai

@lifeizhou-ap lifeizhou-ap merged commit 9aee763 into main Jan 7, 2026
20 checks passed
@lifeizhou-ap lifeizhou-ap deleted the lifei/fix-missing-tool-call-copilot-claude branch January 7, 2026 21:23
zanesq added a commit that referenced this pull request Jan 8, 2026
@zanesq
Merge branch 'main' of github.com:block/goose into zane/working-dir
08fd418 
* 'main' of github.com:block/goose:
  Fixed fonts (#6389)
  Update confidence levels prompt injection detection to reduce false positive rates (#6390)
  Add ML-based prompt injection detection  (#5623)
  docs: update custom extensions tutorial (#6388)
  fix ResultsFormat error when loading old sessions (#6385)
  docs: add MCP Apps tutorial and documentation updates (#6384)
  changed z-index to make sure the search highlighter does not appear on modal overlay (#6386)
  Handling special claude model response in github copilot provider (#6369)
  fix: prevent duplicate rendering when tool returns both mcp-ui and mcp-apps resources (#6378)
  fix: update MCP Apps _meta.ui.resourceUri to use nested format (SEP-1865) (#6372)
  feat(providers): add streaming support for Google Gemini provider (#6191)
  Blog: edit links in mcp apps post (#6371)
  fix: prevent infinite loop of tool-input notifications in MCP Apps (#6374)
michaelneale added a commit that referenced this pull request Jan 8, 2026
@michaelneale
Merge branch 'main' into micn/http-proxy
75b60db 
* main: (31 commits)
  added validation and debug for invalid call tool result (#6368)
  Update MCP apps tutorial: fix _meta structure and version prereq (#6404)
  Fixed fonts (#6389)
  Update confidence levels prompt injection detection to reduce false positive rates (#6390)
  Add ML-based prompt injection detection  (#5623)
  docs: update custom extensions tutorial (#6388)
  fix ResultsFormat error when loading old sessions (#6385)
  docs: add MCP Apps tutorial and documentation updates (#6384)
  changed z-index to make sure the search highlighter does not appear on modal overlay (#6386)
  Handling special claude model response in github copilot provider (#6369)
  fix: prevent duplicate rendering when tool returns both mcp-ui and mcp-apps resources (#6378)
  fix: update MCP Apps _meta.ui.resourceUri to use nested format (SEP-1865) (#6372)
  feat(providers): add streaming support for Google Gemini provider (#6191)
  Blog: edit links in mcp apps post (#6371)
  fix: prevent infinite loop of tool-input notifications in MCP Apps (#6374)
  fix: Show platform-specific keyboard shortcuts in UI (#6323)
  fix: we load extensions when agent starts so don't do it up front (#6350)
  docs: credit HumanLayer in RPI tutorial (#6365)
  Blog: Goose Lands MCP Apps (#6172)
  Claude 3.7 is out. we had some harcoded stuff (#6197)
  ...
wpfleger96 added a commit that referenced this pull request Jan 9, 2026
@wpfleger96
Merge branch 'main' into wpfleger/fork-session
b4105b9 
* main: (89 commits)
  fix(google): treat signed text as regular content in streaming (#6400)
  Add frameDomains and baseUriDomains CSP support for MCP Apps (#6399)
  fix(ci): add missing dependencies to openapi-schema-check job (#6367)
  feat: http proxy support
  Add support for changing working dir and extensions in same window/session (#6057)
  Sort keys in canonical models (#6403)
  added validation and debug for invalid call tool result (#6368)
  Update MCP apps tutorial: fix _meta structure and version prereq (#6404)
  Fixed fonts (#6389)
  Update confidence levels prompt injection detection to reduce false positive rates (#6390)
  Add ML-based prompt injection detection  (#5623)
  docs: update custom extensions tutorial (#6388)
  fix ResultsFormat error when loading old sessions (#6385)
  docs: add MCP Apps tutorial and documentation updates (#6384)
  changed z-index to make sure the search highlighter does not appear on modal overlay (#6386)
  Handling special claude model response in github copilot provider (#6369)
  fix: prevent duplicate rendering when tool returns both mcp-ui and mcp-apps resources (#6378)
  fix: update MCP Apps _meta.ui.resourceUri to use nested format (SEP-1865) (#6372)
  feat(providers): add streaming support for Google Gemini provider (#6191)
  Blog: edit links in mcp apps post (#6371)
  ...
wpfleger96 added a commit that referenced this pull request Jan 9, 2026
@wpfleger96
Merge branch 'main' into wpfleger/fork-session
1b9cf6f 
* main: (89 commits)
  fix(google): treat signed text as regular content in streaming (#6400)
  Add frameDomains and baseUriDomains CSP support for MCP Apps (#6399)
  fix(ci): add missing dependencies to openapi-schema-check job (#6367)
  feat: http proxy support
  Add support for changing working dir and extensions in same window/session (#6057)
  Sort keys in canonical models (#6403)
  added validation and debug for invalid call tool result (#6368)
  Update MCP apps tutorial: fix _meta structure and version prereq (#6404)
  Fixed fonts (#6389)
  Update confidence levels prompt injection detection to reduce false positive rates (#6390)
  Add ML-based prompt injection detection  (#5623)
  docs: update custom extensions tutorial (#6388)
  fix ResultsFormat error when loading old sessions (#6385)
  docs: add MCP Apps tutorial and documentation updates (#6384)
  changed z-index to make sure the search highlighter does not appear on modal overlay (#6386)
  Handling special claude model response in github copilot provider (#6369)
  fix: prevent duplicate rendering when tool returns both mcp-ui and mcp-apps resources (#6378)
  fix: update MCP Apps _meta.ui.resourceUri to use nested format (SEP-1865) (#6372)
  feat(providers): add streaming support for Google Gemini provider (#6191)
  Blog: edit links in mcp apps post (#6371)
  ...
@github-actions github-actions bot mentioned this pull request Jan 13, 2026
Merged
fbalicchia pushed a commit to fbalicchia/goose that referenced this pull request Jan 13, 2026
@lifeizhou-ap @fbalicchia
Handling special claude model response in github copilot provider (bl…
08ab993 
…ock#6369)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

Copilot code review Copilot Copilot left review comments

@jamadeo jamadeo jamadeo approved these changes

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@lifeizhou-ap @jamadeo