Disable thinking for tool call labels - #11207
Merged
Merged
Conversation
…l-thinking # Conflicts: # crates/goose/src/tool_call_labels.rs
filipkujawa
approved these changes
Aug 19, 2026
DOsinga
enabled auto-merge
August 19, 2026 22:46
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 14ec632666
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if let Ok((response, _)) = with_session_id( | ||
| Some(session_id.to_string()), | ||
| provider.complete(model_config, system_prompt, from_ref(message), &[]), | ||
| if let Ok((response, _)) = crate::model_config::complete_fast( |
There was a problem hiding this comment.
When the session uses local_inference with its default enable_thinking = true, routing the label through complete_fast still does not disable reasoning. complete_fast only sets thinking_effort=off, while LocalInferenceProvider::stream in crates/goose-local-inference/src/lib.rs:667-677 exclusively checks whether enable_thinking is explicitly false; the registry default remains true (local_model_registry.rs:90-91,140). Consequently these label calls still run the local model's thinking path, so this change does not achieve its latency/token objective for local models; the one-shot config also needs to override enable_thinking.
Useful? React with 👍 / 👎.
…l-thinking # Conflicts: # crates/goose/src/model_config.rs
michaelneale
added a commit
that referenced
this pull request
Aug 20, 2026
* origin/main: fix: bound retry command diagnostics (#11365) Disable thinking for tool call labels (#11207) fix(review): contain REVIEW.md discovery (#11367) fix(acp): preserve tool result audience metadata (#11375) test(providers): isolate environment-proxy test in its own binary (#11262) fix(security): bind Foundry API keys to request origin (#11347) fix: canonicalize mangled tool names before permission inspection (follow-up to #10230) (#10285) feat(providers): add Lynkr as a declarative OpenAI-compatible provider (#11372) fix: sanitize Pi imported output (#10990) fix(otel): honor RUST_LOG directives for logs (#11360) feat(skills): add web-search and browser-use built-in skills (#11233) feat(dictation): add model-native audio transcription provider (#10589) feat(aws_bedrock): route OpenAI GPT-5.6 (sol/terra/luna) via Bedrock … (#10502) refactor(goose-local-inference): move mlx deps under macos (#11328) feat(providers): add PleumRouter declarative provider (#10479)
jbg
added a commit
that referenced
this pull request
Aug 20, 2026
* origin/main: (50 commits) chore(deps): bump the ui-minor-and-patch group across 1 directory with 53 updates (#11386) fix(security): bound call graph traversal (#11193) fix: pin arrayref to known-good commit (#11389) feat(providers): add SayGM as declarative OpenAI-compatible provider (#11267) feat(providers): custom provider cost fields drive cost tracking (config-declared pricing fallback) (#11220) fix(deps): repair dangling syn reference in Cargo.lock (#11385) fix(flake): add cudaforge hash for git dependency (#10910) feat: auto-focus chat input when user starts typing (#11184) fix(security): fail closed on invalid Codex ACP mode (#11362) fix(mcp): keep stdio extensions alive across worker exits (#10364) feat(ui): collapse scheduled job sessions into accordion in chat history (#11265) fix: bound retry command diagnostics (#11365) Disable thinking for tool call labels (#11207) fix(review): contain REVIEW.md discovery (#11367) fix(acp): preserve tool result audience metadata (#11375) test(providers): isolate environment-proxy test in its own binary (#11262) fix(security): bind Foundry API keys to request origin (#11347) fix: canonicalize mangled tool names before permission inspection (follow-up to #10230) (#10285) feat(providers): add Lynkr as a declarative OpenAI-compatible provider (#11372) fix: sanitize Pi imported output (#10990) ... # Conflicts: # crates/goose/src/agents/state_machine/tests/hooks_lifecycle.rs
jbg
added a commit
that referenced
this pull request
Aug 20, 2026
* origin/main: (59 commits) chore(deps): bump the ui-minor-and-patch group across 1 directory with 53 updates (#11386) fix(security): bound call graph traversal (#11193) fix: pin arrayref to known-good commit (#11389) feat(providers): add SayGM as declarative OpenAI-compatible provider (#11267) feat(providers): custom provider cost fields drive cost tracking (config-declared pricing fallback) (#11220) fix(deps): repair dangling syn reference in Cargo.lock (#11385) fix(flake): add cudaforge hash for git dependency (#10910) feat: auto-focus chat input when user starts typing (#11184) fix(security): fail closed on invalid Codex ACP mode (#11362) fix(mcp): keep stdio extensions alive across worker exits (#10364) feat(ui): collapse scheduled job sessions into accordion in chat history (#11265) fix: bound retry command diagnostics (#11365) Disable thinking for tool call labels (#11207) fix(review): contain REVIEW.md discovery (#11367) fix(acp): preserve tool result audience metadata (#11375) test(providers): isolate environment-proxy test in its own binary (#11262) fix(security): bind Foundry API keys to request origin (#11347) fix: canonicalize mangled tool names before permission inspection (follow-up to #10230) (#10285) feat(providers): add Lynkr as a declarative OpenAI-compatible provider (#11372) fix: sanitize Pi imported output (#10990) ...
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes #11154
What changed
complete_fastcomplete_fastcallerWhy
Tool-call labels used
get_fast_modelfollowed byprovider.completedirectly. That bypassed the thinking-off behavior incomplete_fast, so reasoning-capable models could spend tokens deliberating over short labels whose reasoning output was discarded.Verification
cargo fmtgit diff --checkcargo test -p goose tool_call_labels(18 passed)